1
00:00:00,000 --> 00:00:01,470

2
00:00:01,470 --> 00:00:05,390
>> [MUSIC PLAYING]

3
00:00:05,390 --> 00:00:11,270

4
00:00:11,270 --> 00:00:12,990
>> ABBY FICHTNER: Hello, I'm Abby Fichtner.

5
00:00:12,990 --> 00:00:15,739
Most people know me as Hacker
Chick, because I do the Hacker Chick

6
00:00:15,739 --> 00:00:18,310
Blog on how to build better technology.

7
00:00:18,310 --> 00:00:20,660
And I'm also over at
Harvard Innovation Labs.

8
00:00:20,660 --> 00:00:22,890
Do you know the Innovation Lab?

9
00:00:22,890 --> 00:00:24,510
OK, so that is wicked fun.

10
00:00:24,510 --> 00:00:26,960
I'm hacker in residence
there, where my role

11
00:00:26,960 --> 00:00:30,330
is to help students do everything from
hacking on cool side projects, all

12
00:00:30,330 --> 00:00:34,710
the way up to starting tech startups.

13
00:00:34,710 --> 00:00:36,760
>> I am a programmer, so
that's my background.

14
00:00:36,760 --> 00:00:42,059
I kind of got into programming and
startups by an interesting route.

15
00:00:42,059 --> 00:00:44,600
When I was in school, I wanted
to be a management consultant,

16
00:00:44,600 --> 00:00:46,619
because I thought that
would be the shit.

17
00:00:46,619 --> 00:00:48,160
I don't know if that's still a thing.

18
00:00:48,160 --> 00:00:50,150
Do students still want to
be management consultants?

19
00:00:50,150 --> 00:00:51,380
Is that considered really cool?

20
00:00:51,380 --> 00:00:52,740
>> OK, so I thought that was really cool.

21
00:00:52,740 --> 00:00:54,670
I landed a job with one of
the top management consulting

22
00:00:54,670 --> 00:00:55,753
firms right out of school.

23
00:00:55,753 --> 00:00:59,450
I was very excited right up
until I started working there,

24
00:00:59,450 --> 00:01:02,477
and then absolutely hated it.

25
00:01:02,477 --> 00:01:03,560
I didn't like the company.

26
00:01:03,560 --> 00:01:04,160
I didn't like the culture.

27
00:01:04,160 --> 00:01:06,870
>> I didn't like anything about it
except that they very bizarrely

28
00:01:06,870 --> 00:01:10,010
put me in programming, which was
really weird, because my title wasn't

29
00:01:10,010 --> 00:01:10,889
programmer.

30
00:01:10,889 --> 00:01:13,430
There was nothing that I can
remember in the interview about,

31
00:01:13,430 --> 00:01:14,820
you're going to be programming.

32
00:01:14,820 --> 00:01:18,100
I thought I was going be consulting
managers, whatever that means.

33
00:01:18,100 --> 00:01:21,270
I'm still not actually sure, but
it made sense to me at the time.

34
00:01:21,270 --> 00:01:23,490
>> So I go there, and they
actually gave me an office,

35
00:01:23,490 --> 00:01:25,270
which was cool, because I think
it's the only job I've ever

36
00:01:25,270 --> 00:01:26,520
had where I had an office.

37
00:01:26,520 --> 00:01:30,530
And they gave me a computer and this big
equipment that the computer was hooked

38
00:01:30,530 --> 00:01:34,770
up to, so I was writing code to control
this equipment, which was really neat.

39
00:01:34,770 --> 00:01:37,370
And that part I actually liked.

40
00:01:37,370 --> 00:01:41,720
>> And I was doing code for the
NSA, which was really weird.

41
00:01:41,720 --> 00:01:44,380
It was my first job out of college.

42
00:01:44,380 --> 00:01:45,660
And so I'm writing this code.

43
00:01:45,660 --> 00:01:47,868
I'm just totally hacking,
because I have no idea what

44
00:01:47,868 --> 00:01:49,920
I'm doing, and trying
to make it do things.

45
00:01:49,920 --> 00:01:54,180
>> And I get to this point where I'm using
libraries to control this equipment.

46
00:01:54,180 --> 00:01:55,950
And I can only do
what's in the libraries,

47
00:01:55,950 --> 00:01:59,675
and the thing that I need to do,
there isn't any functions for.

48
00:01:59,675 --> 00:02:01,537
And I'm like, OK.

49
00:02:01,537 --> 00:02:03,370
But there was a support
number, so I call up

50
00:02:03,370 --> 00:02:06,910
the company who created the software,
and I said I need to do this.

51
00:02:06,910 --> 00:02:10,490
And they were like,
yeah, you can't do that.

52
00:02:10,490 --> 00:02:14,220
And it was my first job out of
school and my first project,

53
00:02:14,220 --> 00:02:17,270
and I just didn't feel like I could
just go to the boss and be like--

54
00:02:17,270 --> 00:02:19,040
and he did just kind
of put me on my own.

55
00:02:19,040 --> 00:02:20,600
>> I didn't really feel like
I could go to the boss

56
00:02:20,600 --> 00:02:23,560
to be like, oh, go tell the NSA sorry,
we're not going to do this for them,

57
00:02:23,560 --> 00:02:25,115
because the library isn't available.

58
00:02:25,115 --> 00:02:27,104
It just didn't seem acceptable.

59
00:02:27,104 --> 00:02:29,770
And so I kind of stayed up all
night hacking something together,

60
00:02:29,770 --> 00:02:31,140
and I made it work.

61
00:02:31,140 --> 00:02:34,619
>> And it was this turning moment
for me, where it just clicked.

62
00:02:34,619 --> 00:02:36,410
And I realized this is
what I wanted to do.

63
00:02:36,410 --> 00:02:39,480
I thought it was the coolest thing
ever, that I was like I did something

64
00:02:39,480 --> 00:02:42,146
that the creators of the software
thought weren't even possible.

65
00:02:42,146 --> 00:02:45,490
And I was possibly the first
person ever to do this, right?

66
00:02:45,490 --> 00:02:49,200
And it wasn't that big of a thing,
but it was just such a cool idea.

67
00:02:49,200 --> 00:02:51,265
>> And so I left the big
management consulting firm,

68
00:02:51,265 --> 00:02:53,640
and I went to work for startups,
because startups are all

69
00:02:53,640 --> 00:02:55,900
about creating things that
no one's ever created before.

70
00:02:55,900 --> 00:02:58,025
And I thought that was the
most awesome thing ever.

71
00:02:58,025 --> 00:03:01,990
So I did that for a number
of years, kind of built out

72
00:03:01,990 --> 00:03:03,530
the technology for startups.

73
00:03:03,530 --> 00:03:08,530
And then I kind of, as I was
saying before, got into this area

74
00:03:08,530 --> 00:03:11,562
where I'm just going around helping
hackers and tech entrepreneurs who

75
00:03:11,562 --> 00:03:14,020
are building innovative,
disruptive products-- helping them

76
00:03:14,020 --> 00:03:16,246
to do that and find ways
to do that in which they

77
00:03:16,246 --> 00:03:17,790
can be successful in the market.

78
00:03:17,790 --> 00:03:22,300
>> So that's what I want to
talk to you guys about today.

79
00:03:22,300 --> 00:03:26,540
So for me, I think it's a really
exciting time to be in this space

80
00:03:26,540 --> 00:03:29,609
right now, because technology is
expanding at this incredible rate,

81
00:03:29,609 --> 00:03:31,650
and it's making all these
opportunities available

82
00:03:31,650 --> 00:03:32,850
that were never available before.

83
00:03:32,850 --> 00:03:35,210
So I feel like we're back to that
point, where you can create things

84
00:03:35,210 --> 00:03:36,564
that no one ever created before.

85
00:03:36,564 --> 00:03:38,730
And especially, you look
at things like 3D printing.

86
00:03:38,730 --> 00:03:43,810
So people are 3D printing things
like human organs or food.

87
00:03:43,810 --> 00:03:46,140
NASA has started 3D
printing food astronauts,

88
00:03:46,140 --> 00:03:50,950
so this is a 3D printer
with dough and pizza sauce

89
00:03:50,950 --> 00:03:53,940
and cheese as its cartridges,
rather than polymers.

90
00:03:53,940 --> 00:03:54,810
>> And cars.

91
00:03:54,810 --> 00:03:58,640
Urbee 3D printed the world's
cheapest and most fuel efficient car,

92
00:03:58,640 --> 00:04:00,640
and they're about to drive
it across the country

93
00:04:00,640 --> 00:04:04,420
on under 10 gallons of
fuel, which is pretty crazy.

94
00:04:04,420 --> 00:04:07,770
And of course, everything going on
with mobile, and the fact with things

95
00:04:07,770 --> 00:04:11,770
like 3D printing are making creating
physical devices so much cheaper

96
00:04:11,770 --> 00:04:15,173
has led to the internet of things,
which is this notion that hey, why do we

97
00:04:15,173 --> 00:04:17,839
have to have the functionality
in our computers and our tablets?

98
00:04:17,839 --> 00:04:19,470
Why don't we take it out
of those and actually

99
00:04:19,470 --> 00:04:21,269
put it right into the
devices, where we care about.

100
00:04:21,269 --> 00:04:23,170
>> And so we're getting
things like-- David Rose

101
00:04:23,170 --> 00:04:25,470
over at the Media Lab created an
umbrella that tells the weather.

102
00:04:25,470 --> 00:04:28,000
And so you can imagine it in
an umbrella stand by the door.

103
00:04:28,000 --> 00:04:30,916
And as it senses you walk past it,
if it's going to rain, it'll blink,

104
00:04:30,916 --> 00:04:33,430
so you know to take it with you.

105
00:04:33,430 --> 00:04:36,200
Or Valour created a bike that
gives you directions and gives you

106
00:04:36,200 --> 00:04:38,540
all of your riding stats.

107
00:04:38,540 --> 00:04:42,390
>> Or Hapi created a fork that
monitors your eating habits

108
00:04:42,390 --> 00:04:44,660
to help you eat more healthy.

109
00:04:44,660 --> 00:04:50,060
And everything from self-driving
cars to mind-controlled helicopters--

110
00:04:50,060 --> 00:04:51,290
>> [LAUGHS SLIGHTLY]

111
00:04:51,290 --> 00:04:54,767
>> Even things that we thought of as
very low-tech, like reading the news.

112
00:04:54,767 --> 00:04:57,350
Gannett just announced that
they're working on virtual reality

113
00:04:57,350 --> 00:05:00,990
journalism, where you absorb
the news not by reading it,

114
00:05:00,990 --> 00:05:05,179
but by actually experiencing
it and being a part of it.

115
00:05:05,179 --> 00:05:07,970
Or other things that we might think
of as low-tech, like gardening,

116
00:05:07,970 --> 00:05:08,700
because you need to de-stress.

117
00:05:08,700 --> 00:05:10,500
Because I don't know
about you guys, but I

118
00:05:10,500 --> 00:05:13,550
would find living the
news being very stressful.

119
00:05:13,550 --> 00:05:15,420
>> [LAUGHS]

120
00:05:15,420 --> 00:05:15,930

121
00:05:15,930 --> 00:05:20,380
>> A team out of MIT, Grove, has
created a produce appliance

122
00:05:20,380 --> 00:05:23,690
that actually, you can put in your
kitchen to grow fruits and vegetables.

123
00:05:23,690 --> 00:05:27,100
And so it's really cool
looking at all of the startups.

124
00:05:27,100 --> 00:05:29,002
There's just this amazing
number of startups

125
00:05:29,002 --> 00:05:30,960
that are out these days
that are trying to take

126
00:05:30,960 --> 00:05:32,293
advantage of these technologies.

127
00:05:32,293 --> 00:05:35,836
And what's really interesting-- just
looking at all of these things that are

128
00:05:35,836 --> 00:05:38,960
coming up, but realizing only a very
small percentage of those startups are

129
00:05:38,960 --> 00:05:40,970
actually going to make
it into the future,

130
00:05:40,970 --> 00:05:46,070
and kind of understanding why some of
them make it and some of them don't.

131
00:05:46,070 --> 00:05:50,124
>> So I gave a talk last month
at an engineering conference,

132
00:05:50,124 --> 00:05:52,040
and I wanted to talk to
them about this topic.

133
00:05:52,040 --> 00:05:53,373
And I thought they're engineers.

134
00:05:53,373 --> 00:05:54,190
They want rules.

135
00:05:54,190 --> 00:05:54,750
Like, I'm an engineer.

136
00:05:54,750 --> 00:05:55,220
I like rules.

137
00:05:55,220 --> 00:05:56,470
It's very nice and neat, right?

138
00:05:56,470 --> 00:05:58,390
So I was trying to come up
with the rules of innovation.

139
00:05:58,390 --> 00:06:00,848
>> And as soon as I did that, I
realized that's kind of silly.

140
00:06:00,848 --> 00:06:05,240
The first rule of innovation is that
there are no rules of innovation.

141
00:06:05,240 --> 00:06:07,160
Because if you're doing
it right, then you're

142
00:06:07,160 --> 00:06:09,237
breaking more rules than your following.

143
00:06:09,237 --> 00:06:12,070
And, of course, Thomas Edison
famously said that "I have not failed.

144
00:06:12,070 --> 00:06:14,736
I've just found 10,000
ways that won't work."

145
00:06:14,736 --> 00:06:17,110
And so, of course, the more
innovative that you're being,

146
00:06:17,110 --> 00:06:18,943
you need to kind of
expect that you're going

147
00:06:18,943 --> 00:06:21,850
to find more ways that don't work.

148
00:06:21,850 --> 00:06:25,320
But the good news is that it's
not a complete black hole.

149
00:06:25,320 --> 00:06:27,780
When you look at the startups
that have been successful,

150
00:06:27,780 --> 00:06:29,515
the innovators that have
built these products that

151
00:06:29,515 --> 00:06:31,470
have been successful in
markets, what you'll see

152
00:06:31,470 --> 00:06:33,770
is time and again, the same
patterns emerging of the things

153
00:06:33,770 --> 00:06:34,640
that they're doing.

154
00:06:34,640 --> 00:06:38,320
And a lot of these, when you
kind of dig down into them,

155
00:06:38,320 --> 00:06:42,070
they're kind of predicated on a lot of
the principles behind Lean and Agile--

156
00:06:42,070 --> 00:06:45,900
and people just taking those and saying,
how can these make sense for a startup?

157
00:06:45,900 --> 00:06:48,160
>> So I want to go through these.

158
00:06:48,160 --> 00:06:50,884
To be honest, I think I'd
like to spend about half

159
00:06:50,884 --> 00:06:52,550
the time on this last one-- this "Focus!

160
00:06:52,550 --> 00:06:53,490
And get shit done."

161
00:06:53,490 --> 00:06:55,839
Because really, that's
what it comes down to.

162
00:06:55,839 --> 00:06:58,380
But I think the first four are
really important to understand

163
00:06:58,380 --> 00:07:00,120
the context and the
mindset that you need

164
00:07:00,120 --> 00:07:03,020
to enter into when you're doing
something really innovative that

165
00:07:03,020 --> 00:07:06,120
hasn't been done before.

166
00:07:06,120 --> 00:07:08,287
>> So the first principle
is eliminate waste,

167
00:07:08,287 --> 00:07:10,370
which, if you know anything
about Lean principles,

168
00:07:10,370 --> 00:07:14,520
that's one of the key
principles of Lean.

169
00:07:14,520 --> 00:07:17,170
And, in fact, Eric Ries, who's
the creator of the Lean startup

170
00:07:17,170 --> 00:07:20,960
methodology, says the number one
most important thing for a startup

171
00:07:20,960 --> 00:07:24,580
is learning to tell the difference
between value and waste--

172
00:07:24,580 --> 00:07:27,310
which is pretty weird, right?

173
00:07:27,310 --> 00:07:30,150
Like how could you not know
what's value and what's waste?

174
00:07:30,150 --> 00:07:33,830
>> But I think it makes more sense if
you think about the roots of Lean.

175
00:07:33,830 --> 00:07:38,560
So Lean comes from Lean manufacturing
Toyota Production System in Japan.

176
00:07:38,560 --> 00:07:43,550
And "waste" is a translation from the
term "muda," which is actually broader.

177
00:07:43,550 --> 00:07:45,860
So really, what you want
to do is eliminated muda.

178
00:07:45,860 --> 00:07:48,260
And muda means not just
anything that's unproductive,

179
00:07:48,260 --> 00:07:50,784
but anything that isn't
adding value today.

180
00:07:50,784 --> 00:07:52,700
Because especially when
you're doing something

181
00:07:52,700 --> 00:07:56,110
so uncertain as doing a startup,
creating something innovative,

182
00:07:56,110 --> 00:07:58,900
if you think that you're
going this way and you

183
00:07:58,900 --> 00:08:01,254
start building something
for this, and then you

184
00:08:01,254 --> 00:08:03,420
find out what's really going
on and you go this way,

185
00:08:03,420 --> 00:08:05,440
then anything you did over
here is wasteful, right?

186
00:08:05,440 --> 00:08:07,110
And so in Agile, we have
an expression called

187
00:08:07,110 --> 00:08:08,710
YAGNI, which is "You
Ain't Gonna Need It."

188
00:08:08,710 --> 00:08:09,340
>> [CHUCKLES]

189
00:08:09,340 --> 00:08:12,630
>> So it's a really good thing to remember
as you're building new technologies.

190
00:08:12,630 --> 00:08:14,713
Anything that you think
that you're going to need,

191
00:08:14,713 --> 00:08:18,510
just assume that you're
not until you do.

192
00:08:18,510 --> 00:08:22,410
>> So it's interesting to look at
examples of startups that have made it

193
00:08:22,410 --> 00:08:24,200
and see where they came from.

194
00:08:24,200 --> 00:08:28,612
So PayPal actually started as a
way to beam payments between PDAs.

195
00:08:28,612 --> 00:08:31,320
But it turned out that the world
wasn't ready for mobile payments

196
00:08:31,320 --> 00:08:32,220
in '99, right?

197
00:08:32,220 --> 00:08:35,919
We're only just starting
to get there now.

198
00:08:35,919 --> 00:08:39,734
>> Flickr started as a massively
multiplayer online role playing game.

199
00:08:39,734 --> 00:08:41,900
But it turned out, like
when people were playing it,

200
00:08:41,900 --> 00:08:43,875
that the most fun aspect
was sharing photos.

201
00:08:43,875 --> 00:08:45,810
It's kind of funny.

202
00:08:45,810 --> 00:08:50,140
>> And then Instagram started
as a gamified Foursquare.

203
00:08:50,140 --> 00:08:53,300
And they actually built out the entire
app and looked at it, and went wow,

204
00:08:53,300 --> 00:08:54,800
there is way too much going on here.

205
00:08:54,800 --> 00:08:55,920
This is way too complex.

206
00:08:55,920 --> 00:08:58,070
And they just scrapped the whole
thing and said, you know what?

207
00:08:58,070 --> 00:09:00,040
We're just going to focus
again on the photos.

208
00:09:00,040 --> 00:09:01,812
And that was what was
successful for them.

209
00:09:01,812 --> 00:09:04,520
And so these are the ones that
made it, but when you kind of look

210
00:09:04,520 --> 00:09:06,760
across the board, the
stats are pretty bleak.

211
00:09:06,760 --> 00:09:09,480
Because the stats are that nine
out of ten new products fail,

212
00:09:09,480 --> 00:09:10,690
which is pretty abysmal.

213
00:09:10,690 --> 00:09:15,070
And as developers, as people
who work with technology,

214
00:09:15,070 --> 00:09:17,100
I think when we look
at a stat like this,

215
00:09:17,100 --> 00:09:19,260
we understand how hard it
is to build the tech when

216
00:09:19,260 --> 00:09:21,551
you're building something
that's not been built before.

217
00:09:21,551 --> 00:09:24,700
And we assume that these are failing
because we can't build the technology.

218
00:09:24,700 --> 00:09:28,780
But when you really dig deep,
what's happening-- these products

219
00:09:28,780 --> 00:09:31,170
aren't failing because the
technology didn't work.

220
00:09:31,170 --> 00:09:33,160
They're failing because
the people who created

221
00:09:33,160 --> 00:09:36,140
them weren't able to
find a market for them.

222
00:09:36,140 --> 00:09:39,510
>> My favorite example of
this is a company called

223
00:09:39,510 --> 00:09:42,250
Actuality Systems, which
was actually here in Boston.

224
00:09:42,250 --> 00:09:44,510
They created a 3D holographic display.

225
00:09:44,510 --> 00:09:46,940
That's pretty badass, right?

226
00:09:46,940 --> 00:09:49,290
They create it, and they
got it working, and then

227
00:09:49,290 --> 00:09:51,530
they spent the next 10
years-- so they created this.

228
00:09:51,530 --> 00:09:52,960
This would be impressive
to create today, right?

229
00:09:52,960 --> 00:09:54,459
They created this over 10 years ago.

230
00:09:54,459 --> 00:09:58,340
They spent the next 10 years trying
unsuccessfully to find a market for it

231
00:09:58,340 --> 00:10:01,340
and create a viable business out of
it, and in the end had to shut down,

232
00:10:01,340 --> 00:10:04,700
and all they could do was sell
off a license for the technology.

233
00:10:04,700 --> 00:10:06,444
So were they successful in innovating?

234
00:10:06,444 --> 00:10:08,110
I mean, they got the technology to work.

235
00:10:08,110 --> 00:10:09,090
That's amazing.

236
00:10:09,090 --> 00:10:12,280
But if you're trying to actually
build a viable business out of this,

237
00:10:12,280 --> 00:10:15,130
not so much.

238
00:10:15,130 --> 00:10:19,210
>> And so what's interesting
is there's been research

239
00:10:19,210 --> 00:10:22,675
into what's the single biggest
predictor of startup failure.

240
00:10:22,675 --> 00:10:26,784
Do any of you want to
guess what this is?

241
00:10:26,784 --> 00:10:29,070
>> AUDIENCE: No market?

242
00:10:29,070 --> 00:10:31,275
>> ABBY FICHTNER: No market, yes.

243
00:10:31,275 --> 00:10:34,400
So something that actually I should
have said-- something that startups do,

244
00:10:34,400 --> 00:10:36,890
that if they do this thing, it's
the biggest predictor that they're

245
00:10:36,890 --> 00:10:38,556
going to fail, or the biggest indicator.

246
00:10:38,556 --> 00:10:41,980

247
00:10:41,980 --> 00:10:46,610
So no market is sort of
something that happens to them.

248
00:10:46,610 --> 00:10:48,480
>> So Don [INAUDIBLE] did
a survey into this,

249
00:10:48,480 --> 00:10:51,355
and what he found was the single
biggest predictor of startup failure

250
00:10:51,355 --> 00:10:57,230
was sticking to the initial
business plan-- which

251
00:10:57,230 --> 00:10:59,630
is pretty confusing, right?

252
00:10:59,630 --> 00:11:03,260
Because if you're starting
on any new venture,

253
00:11:03,260 --> 00:11:05,810
you should try to figure out
if you're on track or not.

254
00:11:05,810 --> 00:11:09,160
Even that terminology, on track, implies
that you're talking according to plan.

255
00:11:09,160 --> 00:11:11,618
And so if sticking to plan
means that you're going to fail,

256
00:11:11,618 --> 00:11:13,240
it's very confusing.

257
00:11:13,240 --> 00:11:14,570
Right?

258
00:11:14,570 --> 00:11:17,260
>> And so that brings us to
innovation pattern number

259
00:11:17,260 --> 00:11:19,800
two, which is that you
should really start small.

260
00:11:19,800 --> 00:11:21,600
And this sort of breaks
our mental model,

261
00:11:21,600 --> 00:11:24,020
I think, for how people think
about how startups operate.

262
00:11:24,020 --> 00:11:27,430
Because I feel like we've got this image
of startups as go big or go home, baby.

263
00:11:27,430 --> 00:11:27,960
Right?

264
00:11:27,960 --> 00:11:29,784
Like I've got a big vision, and boom.

265
00:11:29,784 --> 00:11:32,242
I'm going to go big, and I'm
going to be the next Facebook.

266
00:11:32,242 --> 00:11:34,310
>> But the question is how
do you do that, right?

267
00:11:34,310 --> 00:11:37,950
How do you go from nothing but
an idea to like a billion users,

268
00:11:37,950 --> 00:11:39,910
like Facebook has?

269
00:11:39,910 --> 00:11:42,985
How would you even build out
enough features from day one

270
00:11:42,985 --> 00:11:45,460
that you could appeal
to a billion users?

271
00:11:45,460 --> 00:11:47,980
And even if you wanted to
build the next Facebook

272
00:11:47,980 --> 00:11:50,030
tomorrow, how do you start
getting people on it?

273
00:11:50,030 --> 00:11:54,120
Because would any of you use "the next
Facebook" if no one you knew was on it?

274
00:11:54,120 --> 00:11:55,860
Probably not, right?

275
00:11:55,860 --> 00:12:00,320
>> And so what I view startups
as-- when you're really early

276
00:12:00,320 --> 00:12:04,070
stages-- sort of doing the search for
the intersection of our big vision

277
00:12:04,070 --> 00:12:08,120
of what we want to accomplish with what
reality can actually accommodate today.

278
00:12:08,120 --> 00:12:10,870
And the way that you do this
is usually through a series

279
00:12:10,870 --> 00:12:14,137
of small experiments or small tasks.

280
00:12:14,137 --> 00:12:16,970
So just to take a couple examples
of companies that have made it big

281
00:12:16,970 --> 00:12:19,474
and how they started,
Microsoft started with writing

282
00:12:19,474 --> 00:12:21,640
a version of BASIC, which
is a programming language,

283
00:12:21,640 --> 00:12:24,570
for the Altair, which was
like the first home computer.

284
00:12:24,570 --> 00:12:27,320
So I don't know exactly
how many Altairs were made,

285
00:12:27,320 --> 00:12:29,560
but I'm guessing only a few thousand.

286
00:12:29,560 --> 00:12:31,950
So this is not a big market, right?

287
00:12:31,950 --> 00:12:35,415
>> And then, of course, Facebook, which
is the quintessential-- go big,

288
00:12:35,415 --> 00:12:37,890
become the next Facebook--
started here at Harvard,

289
00:12:37,890 --> 00:12:39,530
where there's only 20,000 students.

290
00:12:39,530 --> 00:12:41,170
So again, not a big market.

291
00:12:41,170 --> 00:12:45,190
And so when you're thinking about
the mental model for how startups

292
00:12:45,190 --> 00:12:47,140
should look, it should
look more like this.

293
00:12:47,140 --> 00:12:50,230
You start with your big
vision, but then you go small.

294
00:12:50,230 --> 00:12:53,040
And you figure out a way to
dominate a really niche market,

295
00:12:53,040 --> 00:12:56,047
and then you can build on
that success to go big.

296
00:12:56,047 --> 00:12:57,630
And there's a couple reasons for this.

297
00:12:57,630 --> 00:13:02,930
One is if we accept the fact that
sticking to the initial business plan's

298
00:13:02,930 --> 00:13:05,600
going to fail, we're going to
find 10,000 ways that don't work,

299
00:13:05,600 --> 00:13:07,840
whatever, we're going to
make a lot of mistakes.

300
00:13:07,840 --> 00:13:10,140
We're going to have a lot of misses.

301
00:13:10,140 --> 00:13:13,140
If we try to go big, we're going to
use up all of our time and resources

302
00:13:13,140 --> 00:13:13,940
on the wrong thing.

303
00:13:13,940 --> 00:13:17,220
And so it's much better to go
small so we can experiment quickly.

304
00:13:17,220 --> 00:13:19,970
>> But even more importantly,
it's so much easier

305
00:13:19,970 --> 00:13:22,819
to be successful when we go
small, because all you have to do

306
00:13:22,819 --> 00:13:25,860
is find that market that you want to
go after-- that really niche market.

307
00:13:25,860 --> 00:13:28,193
And then just identify the
one thing that they're really

308
00:13:28,193 --> 00:13:29,880
dying to have, and build that for them.

309
00:13:29,880 --> 00:13:31,690
And then you can be really compelling.

310
00:13:31,690 --> 00:13:35,782
>> So like the Altair users really wanted
a way to program their computer.

311
00:13:35,782 --> 00:13:38,240
And I don't know-- I think it
was just like toggle switches

312
00:13:38,240 --> 00:13:39,185
and blinking lights, right?

313
00:13:39,185 --> 00:13:40,601
So I don't know how they did that.

314
00:13:40,601 --> 00:13:43,510
So providing BASIC so they
could program it is amazing.

315
00:13:43,510 --> 00:13:46,660
Or Harvard students just wanted a
single, centralized student directory,

316
00:13:46,660 --> 00:13:47,160
right?

317
00:13:47,160 --> 00:13:49,410
And so Facebook only had to
provide that one feature.

318
00:13:49,410 --> 00:13:52,451
They didn't have to build it out like
it is today to really get traction.

319
00:13:52,451 --> 00:13:55,080

320
00:13:55,080 --> 00:13:58,210
>> So that takes us to number
three, which is in order

321
00:13:58,210 --> 00:14:01,940
to find that one feature that
your market is really dying for,

322
00:14:01,940 --> 00:14:04,130
you have to really deeply
understand your customers.

323
00:14:04,130 --> 00:14:06,890

324
00:14:06,890 --> 00:14:11,030
And I feel like people underestimate
the importance of this--

325
00:14:11,030 --> 00:14:13,871
especially today, when there's so
many startups that are out there.

326
00:14:13,871 --> 00:14:16,620
If you are really looking at what's
going on in the startup space,

327
00:14:16,620 --> 00:14:18,950
you're going to find 100 startups
all doing the same thing.

328
00:14:18,950 --> 00:14:19,210
Right?

329
00:14:19,210 --> 00:14:22,140
>> And that's because everybody can see
that technology is here today, right?

330
00:14:22,140 --> 00:14:23,200
But we want to be here.

331
00:14:23,200 --> 00:14:25,780
So people see those gaps, and
everybody tries to go after those gaps.

332
00:14:25,780 --> 00:14:27,260
And you have all these startups
all doing the same thing,

333
00:14:27,260 --> 00:14:29,343
and you're like, why isn't
any of them succeeding?

334
00:14:29,343 --> 00:14:30,479
There's a gap here.

335
00:14:30,479 --> 00:14:32,520
I believe that the ones
that are going to succeed

336
00:14:32,520 --> 00:14:36,340
are the ones that take the time to
really understand their customers.

337
00:14:36,340 --> 00:14:38,820
A great example of this,
I think, is Dropbox.

338
00:14:38,820 --> 00:14:42,280
When Drew Houston, the founder, went
to try to raise money for Dropbox,

339
00:14:42,280 --> 00:14:43,826
the VCs really discouraged him.

340
00:14:43,826 --> 00:14:46,700
They're like, I don't understand
why you're even entering this space.

341
00:14:46,700 --> 00:14:50,490
There's already like a million billion
cloud storage startups out there.

342
00:14:50,490 --> 00:14:54,000
>> And Drew was like, yeah,
but do you use any of them?

343
00:14:54,000 --> 00:14:54,730
And they weren't.

344
00:14:54,730 --> 00:14:57,120
And so I feel like Drew
succeeded because A,

345
00:14:57,120 --> 00:14:58,520
he started with a small market.

346
00:14:58,520 --> 00:14:59,730
He didn't try to go after everybody.

347
00:14:59,730 --> 00:15:01,521
He went after the
hardcore techies who have

348
00:15:01,521 --> 00:15:03,440
a lot of devices, a lot
of computers, and they

349
00:15:03,440 --> 00:15:05,040
have this problem in transferring files.

350
00:15:05,040 --> 00:15:06,123
And he just targeted them.

351
00:15:06,123 --> 00:15:09,780
And all he had to do was provide
a solution that worked for them.

352
00:15:09,780 --> 00:15:13,110
>> So again, I feel like there's
a lot of myths around startups,

353
00:15:13,110 --> 00:15:15,770
because we see so many
startups happening today.

354
00:15:15,770 --> 00:15:19,010
And you just hear the 20,000 foot
view of oh, they made it overnight.

355
00:15:19,010 --> 00:15:21,130
They were a success.

356
00:15:21,130 --> 00:15:24,919
But the myth of if you build it,
they will come-- when you really

357
00:15:24,919 --> 00:15:27,460
dig deep into what's happening
in those success stories, time

358
00:15:27,460 --> 00:15:29,460
and again, I think what
you'll find are founders

359
00:15:29,460 --> 00:15:32,795
who went to these extraordinary
lengths to understand their customers.

360
00:15:32,795 --> 00:15:35,920
So just to give a couple examples-- I
don't know if this is still the case,

361
00:15:35,920 --> 00:15:39,100
but at least initially, one
of the co-founders of Airbnb

362
00:15:39,100 --> 00:15:40,320
did not own or rent a home.

363
00:15:40,320 --> 00:15:42,800
He just went around
and lived in Airbnbs.

364
00:15:42,800 --> 00:15:46,610
Like I don't even know what that looked
like-- like living out of a suitcase?

365
00:15:46,610 --> 00:15:49,000
>> Or Ben Silverman from
Pinterest is amazing at this.

366
00:15:49,000 --> 00:15:53,135
He went and personally reached
out to the first 5,000 customers.

367
00:15:53,135 --> 00:15:54,260
He gave them his cellphone.

368
00:15:54,260 --> 00:15:55,780
He met them for breakfast.

369
00:15:55,780 --> 00:15:58,500
I just spoke to their
CTO a couple weeks ago.

370
00:15:58,500 --> 00:16:00,360
And they're entering
into new countries now,

371
00:16:00,360 --> 00:16:01,890
and he's going out and doing it again.

372
00:16:01,890 --> 00:16:04,723
So he's incredible for going out
and individually talking to people.

373
00:16:04,723 --> 00:16:06,612

374
00:16:06,612 --> 00:16:09,570
So, of course, as you're going out
and having these conversations, what

375
00:16:09,570 --> 00:16:12,069
you want to be doing is always
learning from your customer

376
00:16:12,069 --> 00:16:14,860
about what's going to make sense
and what's going to be successful.

377
00:16:14,860 --> 00:16:17,410

378
00:16:17,410 --> 00:16:19,910
I feel like the best
startups, the best innovators,

379
00:16:19,910 --> 00:16:23,410
treat innovation as if it was a
science experiment-- or in a very

380
00:16:23,410 --> 00:16:25,740
scientific way, I guess I should say.

381
00:16:25,740 --> 00:16:29,000
>> So I'm not a scientist, but as
I understand, scientists come up

382
00:16:29,000 --> 00:16:33,120
with hypotheses, and then they develop
experiments to validate or invalidate

383
00:16:33,120 --> 00:16:34,114
their hypotheses.

384
00:16:34,114 --> 00:16:36,530
And so the question is how can
we do that with innovation?

385
00:16:36,530 --> 00:16:38,340
We have an idea, but it's just an idea.

386
00:16:38,340 --> 00:16:41,051
If we're truly doing something
that's never been done before,

387
00:16:41,051 --> 00:16:42,050
all we have are guesses.

388
00:16:42,050 --> 00:16:42,610
Right?

389
00:16:42,610 --> 00:16:46,290
And so what are some experiments that
we can do to validate or invalidate

390
00:16:46,290 --> 00:16:50,060
those ideas without building
out the entire thing?

391
00:16:50,060 --> 00:16:53,710
>> So talking is great,
and I can't actually

392
00:16:53,710 --> 00:16:55,770
emphasize how strongly--
how important it

393
00:16:55,770 --> 00:16:59,115
is to go out and talk to your
customers, at least initially,

394
00:16:59,115 --> 00:17:01,679
to understand who they
are, what problems they

395
00:17:01,679 --> 00:17:03,470
have today, how they're
solving them today.

396
00:17:03,470 --> 00:17:05,349
But talking can only take you so far.

397
00:17:05,349 --> 00:17:05,849
Right?

398
00:17:05,849 --> 00:17:09,030
You can't use talking to say,
hey, I've got this great idea!

399
00:17:09,030 --> 00:17:10,329
Do you want to buy it?

400
00:17:10,329 --> 00:17:12,537
Because they're going to be
like, oh, yeah of course.

401
00:17:12,537 --> 00:17:13,470
That sounds great.

402
00:17:13,470 --> 00:17:16,437
>> Because people want to encourage you.

403
00:17:16,437 --> 00:17:19,520
They see that you're excited about
something, so they're going to say yes.

404
00:17:19,520 --> 00:17:22,890
And people-- human beings are just
terrible at predicting their behavior.

405
00:17:22,890 --> 00:17:26,650
And so if you ask them-- if you say,
I'm going to, at some point in a future,

406
00:17:26,650 --> 00:17:31,455
release this abstract, hypothetical
product, are you going to want it?

407
00:17:31,455 --> 00:17:34,080
They might say no, but if you
actually put it in front of them,

408
00:17:34,080 --> 00:17:34,630
they might want it.

409
00:17:34,630 --> 00:17:36,700
>> And so really, to do the
test of understanding

410
00:17:36,700 --> 00:17:38,110
if people are going to
want it or not, you really

411
00:17:38,110 --> 00:17:39,820
need to put something in front of them.

412
00:17:39,820 --> 00:17:43,330
So I like this quote from Linus
Torvalds, which is "Talk is cheap.

413
00:17:43,330 --> 00:17:45,060
Show me the code."

414
00:17:45,060 --> 00:17:47,310
Or if you're a startup, you
might say, "Talk is cheap.

415
00:17:47,310 --> 00:17:49,170
Show me the MVP."

416
00:17:49,170 --> 00:17:52,210
>> So have you guys heard MVP,
Minimum Viable Product?

417
00:17:52,210 --> 00:17:55,250
It's kind of this buzzword that
I love and hate at the same time.

418
00:17:55,250 --> 00:17:59,050
Because I love the concept of it,
but it gets a little bit overused.

419
00:17:59,050 --> 00:18:02,310
But the idea is valid,
which is don't go build out

420
00:18:02,310 --> 00:18:04,700
this product that's going
to take you a year to build.

421
00:18:04,700 --> 00:18:08,260
Instead, figure out what's that one
thing that people are dying for?

422
00:18:08,260 --> 00:18:10,310
What's the minimum thing
I can build for them?

423
00:18:10,310 --> 00:18:14,500
And put that in front of
them, and see how they react.

424
00:18:14,500 --> 00:18:18,755
>> So quintessential MVP is a landing page.

425
00:18:18,755 --> 00:18:20,130
I'm sure you guys have seen this.

426
00:18:20,130 --> 00:18:25,257
If you tried to sign up for Ello or
Gmail's new inbox, and they're like oh,

427
00:18:25,257 --> 00:18:26,090
we're not ready yet!

428
00:18:26,090 --> 00:18:28,040
I guess those are a little
different, because those are ready.

429
00:18:28,040 --> 00:18:31,180
But they give you a landing page, and
it's like, it's invite only right now.

430
00:18:31,180 --> 00:18:32,490
But give us your email address.

431
00:18:32,490 --> 00:18:33,512
Right

432
00:18:33,512 --> 00:18:36,470
A lot of places will do this before
they've even built out the product,

433
00:18:36,470 --> 00:18:38,094
just to see if there's interest or not.

434
00:18:38,094 --> 00:18:41,420
So with Dropbox, drew Houston, there
was complex technology behind it.

435
00:18:41,420 --> 00:18:44,420
So he went, and he figured out the
technology-- kind of proved that out,

436
00:18:44,420 --> 00:18:45,586
that that was going to work.

437
00:18:45,586 --> 00:18:48,040
But before he built
out the final product,

438
00:18:48,040 --> 00:18:51,730
he did this mock-up on his computer,
this three-minute screencast

439
00:18:51,730 --> 00:18:54,497
video-- very scrappy.

440
00:18:54,497 --> 00:18:57,330
Put it out on Hacker News, because
he knew was sort of his audience,

441
00:18:57,330 --> 00:18:58,950
were the really technical people.

442
00:18:58,950 --> 00:19:01,430
Put up a landing page that
just said, here's the video.

443
00:19:01,430 --> 00:19:04,850
We haven't launched yet, but if you're
interested, give us your email address.

444
00:19:04,850 --> 00:19:08,642
>> Overnight, got 75,000
sign-ups, which is incredible.

445
00:19:08,642 --> 00:19:10,850
Even today, that would be
impressive, but today, they

446
00:19:10,850 --> 00:19:12,520
have like 300 million users, right?

447
00:19:12,520 --> 00:19:14,450
When he posted this,
nobody knew who Dropbox

448
00:19:14,450 --> 00:19:16,320
was because they didn't exist yet.

449
00:19:16,320 --> 00:19:19,420
And so that was a really strong signal
that he had gotten something right.

450
00:19:19,420 --> 00:19:22,490

451
00:19:22,490 --> 00:19:27,260
>> To give you a little bit more
extensive of an example of that,

452
00:19:27,260 --> 00:19:29,800
do you guys know Buffer?

453
00:19:29,800 --> 00:19:32,350
It's a social media
sharing site, and the idea

454
00:19:32,350 --> 00:19:36,440
is-- I tend to read news
at like 2:00 AM, because I

455
00:19:36,440 --> 00:19:37,620
don't want to go to sleep.

456
00:19:37,620 --> 00:19:39,540
And so I might read like 10
articles that are all really cool

457
00:19:39,540 --> 00:19:41,081
and I want to share them with people.

458
00:19:41,081 --> 00:19:43,630
But A, if I share them
out on Twitter right now,

459
00:19:43,630 --> 00:19:45,370
nobody is awake at
2:00 AM except for me.

460
00:19:45,370 --> 00:19:47,453
And B, if they are awake,
they're like why are you

461
00:19:47,453 --> 00:19:49,489
spamming me with 10
articles at once, right?

462
00:19:49,489 --> 00:19:51,780
And so what it does is it's
kind of a queue or a buffer

463
00:19:51,780 --> 00:19:56,330
that you add things to, and it'll
push them out a couple times a day

464
00:19:56,330 --> 00:19:58,570
at a more realistic schedule.

465
00:19:58,570 --> 00:20:00,050
>> So this is how it looks today.

466
00:20:00,050 --> 00:20:02,030
That's not how it started.

467
00:20:02,030 --> 00:20:05,639
The founder had this idea, and
he thought this was a good idea,

468
00:20:05,639 --> 00:20:06,930
but he didn't want to build it.

469
00:20:06,930 --> 00:20:08,964
He didn't want to quit
his day job yet until he

470
00:20:08,964 --> 00:20:11,880
got some validation that other people
thought it was a good idea, too.

471
00:20:11,880 --> 00:20:13,342
So he didn't even need a video.

472
00:20:13,342 --> 00:20:14,550
It was such a simple concept.

473
00:20:14,550 --> 00:20:17,630
>> Just start with Twitter,
puts up a landing page.

474
00:20:17,630 --> 00:20:18,590
This is what we do.

475
00:20:18,590 --> 00:20:19,540
He tweets it out.

476
00:20:19,540 --> 00:20:23,755
When people click Plans and
Pricing, it just gives them

477
00:20:23,755 --> 00:20:26,380
a "you caught us before we're
ready." but if you're interested,

478
00:20:26,380 --> 00:20:27,810
give us your email address.

479
00:20:27,810 --> 00:20:28,600
Tweets it out.

480
00:20:28,600 --> 00:20:30,460
People went to the site.

481
00:20:30,460 --> 00:20:31,960
They were given their email address.

482
00:20:31,960 --> 00:20:35,250
>> He was like, OK, that's a pretty good
indicator that there's some interest,

483
00:20:35,250 --> 00:20:36,920
so I'm ready to go to the next step.

484
00:20:36,920 --> 00:20:38,300
But I don't want to build it yet.

485
00:20:38,300 --> 00:20:41,216
I want to see-- people are interested,
but can I make money off of it?

486
00:20:41,216 --> 00:20:42,570
Can I make it into a business?

487
00:20:42,570 --> 00:20:47,450
So all he did was added a middle page
when people clicked Plans and Pricing

488
00:20:47,450 --> 00:20:49,160
with three pricing plans-- one was free.

489
00:20:49,160 --> 00:20:50,640
Two were paid.

490
00:20:50,640 --> 00:20:51,980
>> Kept tweeting it out.

491
00:20:51,980 --> 00:20:53,892
People kept clicking.

492
00:20:53,892 --> 00:20:56,600
Most people did the free plan,
but some people did the paid plan.

493
00:20:56,600 --> 00:20:57,641
He's like, you know what?

494
00:20:57,641 --> 00:21:00,850
That's enough validation-- not
for me maybe to quit my day job

495
00:21:00,850 --> 00:21:03,770
and spend a year on this, but for
me to just go heads-down and do

496
00:21:03,770 --> 00:21:05,880
a really simple version of this.

497
00:21:05,880 --> 00:21:07,630
He thought it was going
to take him a day.

498
00:21:07,630 --> 00:21:09,713
Technology's hard, so it
took him like seven days.

499
00:21:09,713 --> 00:21:12,620
But it was enough for him
to spend seven days on it.

500
00:21:12,620 --> 00:21:16,097
>> And very quickly, he started
getting users on that first version,

501
00:21:16,097 --> 00:21:17,430
even though it was very minimal.

502
00:21:17,430 --> 00:21:19,180
And what was awesome
about that was he was

503
00:21:19,180 --> 00:21:22,120
able to see how people
were really using it,

504
00:21:22,120 --> 00:21:25,260
and then kind of evolve
it based on them using it.

505
00:21:25,260 --> 00:21:27,760
So Buffer's wonderful, because
it's a really simple example.

506
00:21:27,760 --> 00:21:29,560
Not all technology is
that simple, but this

507
00:21:29,560 --> 00:21:32,250
is sort of the quintessential
Lean startup approach, right?

508
00:21:32,250 --> 00:21:34,260
This is great-- you're
testing it every step,

509
00:21:34,260 --> 00:21:36,257
and you're only going
far enough that you've

510
00:21:36,257 --> 00:21:38,340
validated that it's kind
of worth your time to do.

511
00:21:38,340 --> 00:21:41,330

512
00:21:41,330 --> 00:21:44,170
>> Another great way to get
validation, of course,

513
00:21:44,170 --> 00:21:46,940
is doing a crowdfunding
campaign like Kickstarter,

514
00:21:46,940 --> 00:21:48,630
where you can get pre-orders.

515
00:21:48,630 --> 00:21:51,710
This makes a lot of sense if you're
doing anything that's hardware.

516
00:21:51,710 --> 00:21:54,760
Again, Pebble was the
biggest Kickstarter

517
00:21:54,760 --> 00:22:01,170
until that title got taken by a
cooler-- did you guys see this?

518
00:22:01,170 --> 00:22:04,030
Like an actual cooler that you
bring to the picnic beat out,

519
00:22:04,030 --> 00:22:05,446
so they got more than $10 million.

520
00:22:05,446 --> 00:22:06,998
[LAUGHS SLIGHTLY]

521
00:22:06,998 --> 00:22:08,489

522
00:22:08,489 --> 00:22:11,700
>> But again, like Dropbox, with
Pebble, it was complex technology.

523
00:22:11,700 --> 00:22:14,899
They had to do a proof of concept,
make sure they could prove out

524
00:22:14,899 --> 00:22:16,190
that the technology could work.

525
00:22:16,190 --> 00:22:19,830
But then it's expensive to manufacture,
so before they actually manufactured,

526
00:22:19,830 --> 00:22:21,310
they put up a Kickstarter.

527
00:22:21,310 --> 00:22:24,170
And they used it to
get pre-orders, right?

528
00:22:24,170 --> 00:22:26,570
They said if we can get
$100,000 in pre-orders,

529
00:22:26,570 --> 00:22:28,420
it's worth it to go forward.

530
00:22:28,420 --> 00:22:31,390
They got $10 million, so doing
pretty good-- pretty good validation.

531
00:22:31,390 --> 00:22:33,950

532
00:22:33,950 --> 00:22:39,090
>> So these ideas are all really
great, but as we say in startups,

533
00:22:39,090 --> 00:22:40,159
ideas are a dime a dozen.

534
00:22:40,159 --> 00:22:41,200
It's all about execution.

535
00:22:41,200 --> 00:22:44,300

536
00:22:44,300 --> 00:22:46,310
So this is my favorite
part is the "Focus!

537
00:22:46,310 --> 00:22:49,750
And get shit done."

538
00:22:49,750 --> 00:22:53,110
So the best entrepreneurs
are able to just have

539
00:22:53,110 --> 00:22:58,280
this crazy, intense hyper-focus and
get things done at an amazing pace.

540
00:22:58,280 --> 00:23:03,940
>> So I kind of walk through some
of the development practices.

541
00:23:03,940 --> 00:23:06,330
And ask questions if you have them.

542
00:23:06,330 --> 00:23:09,870
I wasn't quite sure how much you guys
knew about development practices,

543
00:23:09,870 --> 00:23:12,610
so kind of have a
discussion about what that

544
00:23:12,610 --> 00:23:15,890
looks like when you're
developing something like this.

545
00:23:15,890 --> 00:23:17,900
So the first thing is
to figure out OK, what

546
00:23:17,900 --> 00:23:21,205
is it that I should be focusing on--
which can be really challenging when

547
00:23:21,205 --> 00:23:22,330
you're doing something new.

548
00:23:22,330 --> 00:23:23,940
Because everybody has all
these ideas, and there's

549
00:23:23,940 --> 00:23:26,310
so many different directions you can
go, and so many different questions

550
00:23:26,310 --> 00:23:27,480
that you have.

551
00:23:27,480 --> 00:23:29,615
>> So step number one, figure
out what to focus on.

552
00:23:29,615 --> 00:23:32,740
A lot of times, as developers, as people
who are thinking about technology,

553
00:23:32,740 --> 00:23:34,448
we're really thinking
about the products.

554
00:23:34,448 --> 00:23:37,720
We think about things kind of in
this order-- first, can I build it?

555
00:23:37,720 --> 00:23:41,420
Assuming that I can build it, then
can I get people to know about it?

556
00:23:41,420 --> 00:23:44,490
Assuming that I can, can
I make money from it?

557
00:23:44,490 --> 00:23:47,820
>> But if we're trying to
do a viable business,

558
00:23:47,820 --> 00:23:51,620
we might want to be thinking
of those in the opposite order.

559
00:23:51,620 --> 00:23:55,246
The reason is I feel like-- and
I do this myself, so I get it.

560
00:23:55,246 --> 00:23:57,620
I feel like we get very hung
up on this "Can I build it?"

561
00:23:57,620 --> 00:24:01,154
question, because if you're a technology
person-- if you're a developer--

562
00:24:01,154 --> 00:24:02,570
you're really thinking about that.

563
00:24:02,570 --> 00:24:05,580
>> But the truth is usually, when we
come up with an idea for a startup,

564
00:24:05,580 --> 00:24:08,337
we're coming up with it based on
I've seen this technology here

565
00:24:08,337 --> 00:24:10,420
and this technology here
and this technology here,

566
00:24:10,420 --> 00:24:12,190
and if I just combine
them in some new way,

567
00:24:12,190 --> 00:24:13,815
I think it would be really interesting.

568
00:24:13,815 --> 00:24:16,610
Well, if I've already seen the
technology in those places,

569
00:24:16,610 --> 00:24:18,160
you kind of know it exists, right?

570
00:24:18,160 --> 00:24:20,340
>> So sure, do some proof of concepts.

571
00:24:20,340 --> 00:24:22,360
If there's some technical risk in there.

572
00:24:22,360 --> 00:24:24,530
But for the most part, the things
that we're coming up with--

573
00:24:24,530 --> 00:24:26,960
unless we're really awesome and doing
something totally new, in which case,

574
00:24:26,960 --> 00:24:28,530
figure out if you can build it.

575
00:24:28,530 --> 00:24:31,420
But usually, most of the
startups I see, you can build it.

576
00:24:31,420 --> 00:24:34,100
That isn't even a question.

577
00:24:34,100 --> 00:24:37,620
>> So start thinking about
is something that people

578
00:24:37,620 --> 00:24:42,070
are going to be able to pay me for
And then how am I going to reach them?

579
00:24:42,070 --> 00:24:45,560
That's really hard, especially
if you are a technical person,

580
00:24:45,560 --> 00:24:48,000
do you have a way to
reach out to these people

581
00:24:48,000 --> 00:24:51,140
and get them to buy your product?

582
00:24:51,140 --> 00:24:57,344
>> So once you figure out, OK, what's that
question-- kind of always have in mind,

583
00:24:57,344 --> 00:25:00,260
this is the most important question
that I need to be driving towards,

584
00:25:00,260 --> 00:25:02,250
or the most important thing
that I need to be validating.

585
00:25:02,250 --> 00:25:04,670
And then you want to get back to
this notion of eliminating waste.

586
00:25:04,670 --> 00:25:07,030
Just figure out like the
leanest, most efficient way

587
00:25:07,030 --> 00:25:10,110
that you can go about
answering that question.

588
00:25:10,110 --> 00:25:12,690
>> So I talked about
minimum viable product.

589
00:25:12,690 --> 00:25:17,584
I would say get into this mindset
of minimum viable everything--

590
00:25:17,584 --> 00:25:20,500
by which I don't mean that you should
be doing a crappy job at things.

591
00:25:20,500 --> 00:25:22,924
I just mean how can
you cut out the waste?

592
00:25:22,924 --> 00:25:25,090
How do you get just right
to the heart of the matter

593
00:25:25,090 --> 00:25:29,010
and figure out how to validate
things without gold-plating,

594
00:25:29,010 --> 00:25:31,460
without doing more than you need to.

595
00:25:31,460 --> 00:25:35,040
>> So just to give some examples,
I feel like initially, you're

596
00:25:35,040 --> 00:25:36,874
trying to figure out I
have this great idea.

597
00:25:36,874 --> 00:25:38,206
Is anyone even going to want it?

598
00:25:38,206 --> 00:25:41,310
So a really easy way to do that is a
landing page, like we talked about.

599
00:25:41,310 --> 00:25:43,060
You don't have to write
any code for that.

600
00:25:43,060 --> 00:25:45,800
There's tools that do it for you.

601
00:25:45,800 --> 00:25:48,670
>> If you say, OK, I figured that out.

602
00:25:48,670 --> 00:25:52,234
Now I want I'm assuming that--
OK, people seem to want it.

603
00:25:52,234 --> 00:25:53,900
Would they actually pay me money for it?

604
00:25:53,900 --> 00:25:56,569
You can do things like what
Buffer did with the pricing page,

605
00:25:56,569 --> 00:25:58,610
or even better, a Kickstarter
and get pre-orders.

606
00:25:58,610 --> 00:26:00,150
Orders

607
00:26:00,150 --> 00:26:03,400
>> The next thing that I think you're
going to be wanting to look at is-- OK,

608
00:26:03,400 --> 00:26:04,691
it seems like people wanted it.

609
00:26:04,691 --> 00:26:09,070
It seems like people will pay
for it, but especially with apps,

610
00:26:09,070 --> 00:26:10,330
will people actually use It?

611
00:26:10,330 --> 00:26:12,580
So I don't know the stats,
but they're pretty abysmal.

612
00:26:12,580 --> 00:26:15,570
A huge number of apps get
downloaded and then never used.

613
00:26:15,570 --> 00:26:16,920
And that isn't helpful.

614
00:26:16,920 --> 00:26:18,850
That's nice that you got a
lot of people downloading it.

615
00:26:18,850 --> 00:26:21,516
But if it's not used, you're not
going to stick around for long.

616
00:26:21,516 --> 00:26:24,746

617
00:26:24,746 --> 00:26:26,620
When you're thinking
about that first version

618
00:26:26,620 --> 00:26:29,203
that you want to put out there--
your minimum viable product--

619
00:26:29,203 --> 00:26:32,200
think about what is it exactly
that I'm trying to test?

620
00:26:32,200 --> 00:26:35,290
And what can I do that
just figures that out?

621
00:26:35,290 --> 00:26:36,790
I just kind of took a guess at this.

622
00:26:36,790 --> 00:26:39,706
I don't actually know what Buffer's
first version looked like exactly.

623
00:26:39,706 --> 00:26:42,659
But if you think about Buffer-- just
because of this simple example--

624
00:26:42,659 --> 00:26:44,450
you might think this
is what they feel like

625
00:26:44,450 --> 00:26:46,879
as their first minimum viable product.

626
00:26:46,879 --> 00:26:48,670
I need to be able to
create a user account,

627
00:26:48,670 --> 00:26:51,180
obviously, link it to my
social media accounts.

628
00:26:51,180 --> 00:26:53,530
I need to add posts like
tweets into my buffer.

629
00:26:53,530 --> 00:26:54,470
Edit them.

630
00:26:54,470 --> 00:26:55,770
Delete them.

631
00:26:55,770 --> 00:26:58,320
>> Set the time when I
want those to be posted.

632
00:26:58,320 --> 00:27:01,650
Obviously, the software needs
to automatically post to Twitter

633
00:27:01,650 --> 00:27:03,470
or whatever based on that schedule.

634
00:27:03,470 --> 00:27:05,970
And then I should be able to
view a history of my post.

635
00:27:05,970 --> 00:27:08,900
That feels pretty minimal,
pretty basic, right?

636
00:27:08,900 --> 00:27:11,710
>> I always encourage startups--
especially like, this is easy for us,

637
00:27:11,710 --> 00:27:12,800
because it's not our baby.

638
00:27:12,800 --> 00:27:13,300
Right?

639
00:27:13,300 --> 00:27:17,220
Be like, oh, Yeah whatever Look at it
again, and keep saying is there a way

640
00:27:17,220 --> 00:27:19,840
that I can get it
stripped down even more?

641
00:27:19,840 --> 00:27:21,860
>> So what is it we're
trying to figure out?

642
00:27:21,860 --> 00:27:23,860
If we're trying to figure
out if they'll use it,

643
00:27:23,860 --> 00:27:26,943
we're trying to see if are they even
going to post anything to the bumper?

644
00:27:26,943 --> 00:27:31,125
So this feels a little hacky, but if
they haven't posted it to the Buffer

645
00:27:31,125 --> 00:27:33,760
yet, you don't really
need to allow them to edit

646
00:27:33,760 --> 00:27:36,357
or delete or view posts in history.

647
00:27:36,357 --> 00:27:38,690
If you can plant that something
out there really quickly

648
00:27:38,690 --> 00:27:42,517
and see if people can even add
postings to it, once you see that,

649
00:27:42,517 --> 00:27:44,850
you can very quickly start
adding on this functionality.

650
00:27:44,850 --> 00:27:48,120
But just get something out there.

651
00:27:48,120 --> 00:27:50,489
Do you need to allow the user
to set a posting schedule?

652
00:27:50,489 --> 00:27:52,780
Probably not, if they're like
me and they're just like,

653
00:27:52,780 --> 00:27:56,490
I don't want my all my treats going
out at 2:00 AM on Sunday night.

654
00:27:56,490 --> 00:27:58,487
>> You can say these are
the most popular times.

655
00:27:58,487 --> 00:28:00,820
Whatever, we're just going
to post it according to that.

656
00:28:00,820 --> 00:28:02,300
You can probably do that.

657
00:28:02,300 --> 00:28:04,950
And then I kind of made this up, because
I know they only started with Twitter.

658
00:28:04,950 --> 00:28:06,991
But obviously, you can
just pick the social media

659
00:28:06,991 --> 00:28:10,050
network that makes the most
sense and just start with that.

660
00:28:10,050 --> 00:28:12,030
And so now you're down
to four out of 10.

661
00:28:12,030 --> 00:28:15,770
>> And if you can get something
out there, a pet peeve of mine

662
00:28:15,770 --> 00:28:17,870
is that people think and
MVP means crappy product.

663
00:28:17,870 --> 00:28:18,994
And I don't think it needs.

664
00:28:18,994 --> 00:28:23,220
I think you can get something
out there that it's still useful,

665
00:28:23,220 --> 00:28:27,151
but isn't gold plated-- is
just the absolute bear minimum.

666
00:28:27,151 --> 00:28:30,400
And I guess you have to kind of figure
out based on your audience what's going

667
00:28:30,400 --> 00:28:31,610
to make sense or what isn't.

668
00:28:31,610 --> 00:28:34,068
>> But a lot of times you get
something out there more minimal

669
00:28:34,068 --> 00:28:38,610
than you'd think-- just a
test, how people use it.

670
00:28:38,610 --> 00:28:41,330
So as you're building
out these features,

671
00:28:41,330 --> 00:28:43,922
you want to think about what's
the minimum viable process.

672
00:28:43,922 --> 00:28:46,880
And so a lot of times when we think
about really lightweight processes,

673
00:28:46,880 --> 00:28:48,171
we think about agile processes.

674
00:28:48,171 --> 00:28:52,100
We think about lean-- this is a little
bit random-- just some agile and lean

675
00:28:52,100 --> 00:28:54,557
books that I like.

676
00:28:54,557 --> 00:28:56,890
So there's great practices
like from Extreme Programming

677
00:28:56,890 --> 00:28:59,075
and continuous integration,
and refactoring,

678
00:28:59,075 --> 00:29:01,010
which I'll speak to a little bit.

679
00:29:01,010 --> 00:29:04,220
But the thing is, once you start getting
into the Agile and mean practices,

680
00:29:04,220 --> 00:29:07,800
it can very quickly get overwhelming.

681
00:29:07,800 --> 00:29:11,260
And it can wind up begin
real overkill for a startup.

682
00:29:11,260 --> 00:29:14,320
>> So the thing is that
a lot of these books

683
00:29:14,320 --> 00:29:16,490
are talking about how
to do Agile when you're

684
00:29:16,490 --> 00:29:18,310
doing a product for an
established company.

685
00:29:18,310 --> 00:29:18,810
Right?

686
00:29:18,810 --> 00:29:21,870
And you know who the market is, and
you know what your product road map.

687
00:29:21,870 --> 00:29:23,130
And they wind up-- even
though we're supposed

688
00:29:23,130 --> 00:29:25,170
to be light weight-- they wind up
actually being way too heavyweight

689
00:29:25,170 --> 00:29:27,003
for our startup, because
the startup is just

690
00:29:27,003 --> 00:29:29,450
operating at this
completely different level.

691
00:29:29,450 --> 00:29:32,340
So my feel is that when
you're going a startup,

692
00:29:32,340 --> 00:29:33,750
you need to be scrappy as hell.

693
00:29:33,750 --> 00:29:34,300
Right?

694
00:29:34,300 --> 00:29:36,180
So initially, there's no process.

695
00:29:36,180 --> 00:29:37,930
You want to keep it
as simple as possible.

696
00:29:37,930 --> 00:29:41,070
And only add process that's
sort of a just-in-time process.

697
00:29:41,070 --> 00:29:42,490
OK, we see that there's a problem?

698
00:29:42,490 --> 00:29:45,345
Let's add just enough process
to address that problem.

699
00:29:45,345 --> 00:29:47,396
Do you know what I mean?

700
00:29:47,396 --> 00:29:51,970
It's because you don't want any
of us holding you down, right?

701
00:29:51,970 --> 00:29:56,030
>> Scrum is a really popular
process for Agile development.

702
00:29:56,030 --> 00:29:59,248
I don't know if you guys
are familiar with this.

703
00:29:59,248 --> 00:30:00,560
OK, well--

704
00:30:00,560 --> 00:30:01,680
>> [CHUCKLES]

705
00:30:01,680 --> 00:30:04,194
>> It would be just too
overkill for a startup.

706
00:30:04,194 --> 00:30:05,360
So I won't worry about that.

707
00:30:05,360 --> 00:30:09,370
So OK, if you think about what's the
absolute simplest thing that I need.

708
00:30:09,370 --> 00:30:11,540
Well, I need to probably
keep track of what

709
00:30:11,540 --> 00:30:14,460
I'm doing, especially if
there's more than one person,

710
00:30:14,460 --> 00:30:16,375
but even if there's one person.

711
00:30:16,375 --> 00:30:17,250
What am I working on?

712
00:30:17,250 --> 00:30:20,377
>> So a simple task board-- very easy.

713
00:30:20,377 --> 00:30:21,460
This is what I want to do.

714
00:30:21,460 --> 00:30:22,250
This is what I'm working on.

715
00:30:22,250 --> 00:30:23,410
This is what I've done.

716
00:30:23,410 --> 00:30:28,050
The only problem that I see when I see
startups doing something like this,

717
00:30:28,050 --> 00:30:31,620
is that very quickly,
their in-progress column

718
00:30:31,620 --> 00:30:36,170
tends to look like that, which is not
very helpful-- especially if there's

719
00:30:36,170 --> 00:30:37,990
only one person or only one developer.

720
00:30:37,990 --> 00:30:38,230
Right?

721
00:30:38,230 --> 00:30:39,938
>> Because you're not
getting anything done.

722
00:30:39,938 --> 00:30:44,805
All you're doing is going back and forth
trying to get all these things done.

723
00:30:44,805 --> 00:30:47,930
And so this is a really good example
of where just enough process can come.

724
00:30:47,930 --> 00:30:51,720
So Kanban is a really great tool.

725
00:30:51,720 --> 00:30:53,940
It comes also from Lean manufacturing.

726
00:30:53,940 --> 00:30:57,580
>> And the idea is that what we
want to do is put constraints

727
00:30:57,580 --> 00:31:00,760
around how much work we can
handle at any given time.

728
00:31:00,760 --> 00:31:05,510
And so if we're one person, then we
can only work on one item at a time.

729
00:31:05,510 --> 00:31:06,360
Sorry.

730
00:31:06,360 --> 00:31:08,630
So all that other stuff
needs to go over there.

731
00:31:08,630 --> 00:31:11,310
So what we do is we put work in
progress limits on the columns.

732
00:31:11,310 --> 00:31:12,851
If there's two people, it can be two.

733
00:31:12,851 --> 00:31:16,510
You can figure out what
makes the most sense for you.

734
00:31:16,510 --> 00:31:19,367
>> But the idea is keep things
sane, so that you're just

735
00:31:19,367 --> 00:31:20,450
doing one thing at a time.

736
00:31:20,450 --> 00:31:21,325
You're able to do it.

737
00:31:21,325 --> 00:31:23,130
You're able to actually get it done.

738
00:31:23,130 --> 00:31:27,349
One thing to keep in mind
is-- if you have a one item

739
00:31:27,349 --> 00:31:29,640
that you're doing but the
item takes three months, that

740
00:31:29,640 --> 00:31:31,550
would be a difficult for
a startup, obviously.

741
00:31:31,550 --> 00:31:32,850
You need to be able
to be flexible and be

742
00:31:32,850 --> 00:31:34,340
able to handle things
as they come at you.

743
00:31:34,340 --> 00:31:36,548
You can't say I'm not doing
anything for three months

744
00:31:36,548 --> 00:31:38,450
until I get the login screen done.

745
00:31:38,450 --> 00:31:39,370
I don't know.

746
00:31:39,370 --> 00:31:42,840
>> So I advise startups to
keep this really short,

747
00:31:42,840 --> 00:31:45,180
to keep these tasks so
that they fit into a day.

748
00:31:45,180 --> 00:31:48,305
Obviously, if it's more complex, that
might need to be a little bit longer.

749
00:31:48,305 --> 00:31:49,970
But figure out what works best for you.

750
00:31:49,970 --> 00:31:51,440
You can try different lengths.

751
00:31:51,440 --> 00:31:54,290
But generally, just as an
example, if you keep all the tasks

752
00:31:54,290 --> 00:31:57,640
so they fit within one day, that
means that every single day,

753
00:31:57,640 --> 00:31:58,890
you're getting something done.

754
00:31:58,890 --> 00:32:00,014
And you're providing value.

755
00:32:00,014 --> 00:32:03,300
And that momentum can
really move you forward

756
00:32:03,300 --> 00:32:08,050
instead of the situation before,
where you have 500 things going,

757
00:32:08,050 --> 00:32:10,622
and none of them are done.

758
00:32:10,622 --> 00:32:12,330
The other thing,
though, is still looking

759
00:32:12,330 --> 00:32:15,120
at this to-do column-- I'm
overwhelmed looking at that.

760
00:32:15,120 --> 00:32:18,950
And so if I was a developer and I was
working on A, and I was like oh, shit.

761
00:32:18,950 --> 00:32:21,730
I've got B and C and De and
E and F and G and H. Blah!

762
00:32:21,730 --> 00:32:22,650
Coming down the road.

763
00:32:22,650 --> 00:32:25,060
I'm like freaking out, and I"m trying
to figure out how the design is going

764
00:32:25,060 --> 00:32:26,393
to accommodate all these things.

765
00:32:26,393 --> 00:32:30,004
And the truth is that if we accept the
fact that we don't actually quite know

766
00:32:30,004 --> 00:32:32,920
what the product's going to need to
look like until we've put in front

767
00:32:32,920 --> 00:32:37,030
of a customer, then do we really know
that we need all those tasks yet?

768
00:32:37,030 --> 00:32:38,810
Or are we kind of fooling ourselves?

769
00:32:38,810 --> 00:32:40,890
>> So if you really have
all those ideas, great.

770
00:32:40,890 --> 00:32:43,515
Put them in a notebook or a
spreadsheet or something like that.

771
00:32:43,515 --> 00:32:46,280
But I advise startups to
keep a work-in-progress limit

772
00:32:46,280 --> 00:32:47,360
on the to-do column, too.

773
00:32:47,360 --> 00:32:50,010

774
00:32:50,010 --> 00:32:52,430
That's an absolute maximum,
I would say, how much

775
00:32:52,430 --> 00:32:54,500
you can get done in one or two weeks.

776
00:32:54,500 --> 00:32:56,950
So it doesn't even have to be that many.

777
00:32:56,950 --> 00:32:59,591
>> That way you are just
hyper-focused on this

778
00:32:59,591 --> 00:33:01,340
is what I'm doing,
getting done this week.

779
00:33:01,340 --> 00:33:03,140
Or maybe these two weeks, right?

780
00:33:03,140 --> 00:33:04,935
And nothing else is getting
in your way, and you're just

781
00:33:04,935 --> 00:33:06,893
making sure that you're
getting that out there.

782
00:33:06,893 --> 00:33:10,932
And especially as you start adding
new team members, this really helps.

783
00:33:10,932 --> 00:33:13,390
A lot of people like to do this
in software, which you can.

784
00:33:13,390 --> 00:33:15,740
But it's even better if you
all can be in the same space

785
00:33:15,740 --> 00:33:16,740
and just put it up on a wall.

786
00:33:16,740 --> 00:33:18,060
It's just really visible,
and everyone can just

787
00:33:18,060 --> 00:33:19,643
see it, and see what's most important.

788
00:33:19,643 --> 00:33:22,310

789
00:33:22,310 --> 00:33:24,860
>> So OK, that's how you're
figuring out what to do.

790
00:33:24,860 --> 00:33:26,060
As you're doing it,
you want to be thinking

791
00:33:26,060 --> 00:33:27,684
about what's the minimum viable design?

792
00:33:27,684 --> 00:33:31,040
Or in Agile, we actually
have something called

793
00:33:31,040 --> 00:33:33,180
emergent design, which is the same idea.

794
00:33:33,180 --> 00:33:36,190
So have you guys heard of
emergent design before?

795
00:33:36,190 --> 00:33:36,850
OK.

796
00:33:36,850 --> 00:33:42,060
>> S-- actually, I'm trying
to remember where-- OK.

797
00:33:42,060 --> 00:33:45,037
So the idea of a merchant
design is rather than coming up

798
00:33:45,037 --> 00:33:46,870
with this big, upfront
design and saying I'm

799
00:33:46,870 --> 00:33:49,960
going to spend a month figuring out
the right architecture what components

800
00:33:49,960 --> 00:33:54,280
go where and everything, let me
just design enough for the features

801
00:33:54,280 --> 00:33:56,220
that I know I'm putting
in this first release.

802
00:33:56,220 --> 00:33:59,510
And nothing else-- or the features
that I'm doing this week, even.

803
00:33:59,510 --> 00:34:03,790
>> And then only as I need new features
do I figure out the design for those.

804
00:34:03,790 --> 00:34:06,280
You're not figuring out design upfront.

805
00:34:06,280 --> 00:34:11,650
I think in reality, it's not this
on-off switch or this toggle.

806
00:34:11,650 --> 00:34:14,780
I think it's more of a
spectrum of where you've fall

807
00:34:14,780 --> 00:34:16,241
on the certainty to uncertainty.

808
00:34:16,241 --> 00:34:19,199
And so if in a startup up, or if
you're building something that's never

809
00:34:19,199 --> 00:34:21,948
been built before, you're pretty
far over on the uncertainty curve

810
00:34:21,948 --> 00:34:22,495
here, right?

811
00:34:22,495 --> 00:34:25,670
>> And if you think about it in
terms of the business plan-- like,

812
00:34:25,670 --> 00:34:28,272
we talked about the single
biggest predictor of failure

813
00:34:28,272 --> 00:34:29,980
is sticking to the
initial business plan.

814
00:34:29,980 --> 00:34:32,033
If you do this big
upfront business plan,

815
00:34:32,033 --> 00:34:34,949
and you say I'm just going to blindly
follow that and not do anything.

816
00:34:34,949 --> 00:34:36,989
But you're just going to fail, right?

817
00:34:36,989 --> 00:34:38,659
Because there was too much uncertainty.

818
00:34:38,659 --> 00:34:40,960
And I feel like the
same is true for design.

819
00:34:40,960 --> 00:34:44,790

820
00:34:44,790 --> 00:34:48,181
>> Sorry, so instead of doing
a big upfront business plan,

821
00:34:48,181 --> 00:34:50,139
you would do a very light
weight business model

822
00:34:50,139 --> 00:34:51,750
canvas, which you might have heard of.

823
00:34:51,750 --> 00:34:53,239
It's like a one-pager,
just getting my ideas out.

824
00:34:53,239 --> 00:34:54,679
It's not that you don't
think about it at all.

825
00:34:54,679 --> 00:34:56,110
It's good to think about it at first.

826
00:34:56,110 --> 00:34:58,960
But just get it something really
flexible out there-- just one page.

827
00:34:58,960 --> 00:35:02,979
And then, as you go, kind of
emerge that plan over time

828
00:35:02,979 --> 00:35:05,270
as you learn from customers,
and you can adapt to them.

829
00:35:05,270 --> 00:35:07,450
>> And so then the same
thing is true for design.

830
00:35:07,450 --> 00:35:09,200
You can do a big,
upfront design, but that

831
00:35:09,200 --> 00:35:11,720
doesn't make sense if
there's a lot of uncertainty.

832
00:35:11,720 --> 00:35:15,080
A lot of people would argue there's
never that much certainty in software,

833
00:35:15,080 --> 00:35:16,830
even if you're not doing in startup.

834
00:35:16,830 --> 00:35:19,290
So you never want to do that
big of an upfront design.

835
00:35:19,290 --> 00:35:23,770
But I feel like the
level of design is going

836
00:35:23,770 --> 00:35:27,207
to vary based on how much
certainty or uncertainty there is.

837
00:35:27,207 --> 00:35:30,290
And so if you have no freaking clue
and you're just throwing something out

838
00:35:30,290 --> 00:35:32,124
there like a landing
page, obviously, you're

839
00:35:32,124 --> 00:35:34,539
not going to go take the time
to architect a whole system.

840
00:35:34,539 --> 00:35:35,590
That's ridiculous, right?

841
00:35:35,590 --> 00:35:37,131
So you don't need any upfront design.

842
00:35:37,131 --> 00:35:41,890
A lot of times, the first version
you put out of software for a startup

843
00:35:41,890 --> 00:35:42,890
just gets thrown away.

844
00:35:42,890 --> 00:35:45,056
And so a lot of times, even
though I might say this,

845
00:35:45,056 --> 00:35:46,970
you can just kind of
hack something together.

846
00:35:46,970 --> 00:35:48,553
It's probably going to be thrown away.

847
00:35:48,553 --> 00:35:52,680
But again, use that just-in-time
idea for design as well.

848
00:35:52,680 --> 00:35:54,099
That OK, you know what?

849
00:35:54,099 --> 00:35:55,390
This is actually some traction.

850
00:35:55,390 --> 00:35:58,204
Some people are interested in this.

851
00:35:58,204 --> 00:35:59,620
I'm going to add some features on.

852
00:35:59,620 --> 00:36:03,060
Now, I feel like I should be a
little bit smarter about the design.

853
00:36:03,060 --> 00:36:06,910
>> So the idea is as your designing,
just keep this YAGNI in mind.

854
00:36:06,910 --> 00:36:07,910
You Ain't Gonna Need It.

855
00:36:07,910 --> 00:36:09,830
Don't design for things
that aren't there yet.

856
00:36:09,830 --> 00:36:14,260
And the keep it simple,
stupid principle--

857
00:36:14,260 --> 00:36:16,840
do the simplest thing
that could possibly work.

858
00:36:16,840 --> 00:36:21,090
>> A lot of times, it's interesting,
because as developers,

859
00:36:21,090 --> 00:36:23,480
we get taught to do these
really complex designs.

860
00:36:23,480 --> 00:36:25,513
And we're taught that that's good.

861
00:36:25,513 --> 00:36:28,702
But it prevents us from being
flexible, and it can be really wasteful

862
00:36:28,702 --> 00:36:30,660
if we wind up going in
at different directions.

863
00:36:30,660 --> 00:36:33,310
So Agile kind of says, don't do that.

864
00:36:33,310 --> 00:36:35,950
Just figure out what the
simplest way, the simplest code

865
00:36:35,950 --> 00:36:38,200
that you can put in here
that's going to make it work.

866
00:36:38,200 --> 00:36:42,950
And then if I need to add onto
it, I can kind of fix that code up

867
00:36:42,950 --> 00:36:44,050
and readdress the design.

868
00:36:44,050 --> 00:36:46,883
>> So there's something called refactoring
that's really important when

869
00:36:46,883 --> 00:36:48,450
you do emergent design.

870
00:36:48,450 --> 00:36:56,460
And the idea with refactoring is--
sorry, I'm going to back up a little.

871
00:36:56,460 --> 00:36:59,460
So if you're doing emergent design,
you're only designing for the future

872
00:36:59,460 --> 00:37:00,400
that you have today.

873
00:37:00,400 --> 00:37:02,150
But that doesn't mean
that you're hacking.

874
00:37:02,150 --> 00:37:03,680
That doesn't mean when
you add another feature,

875
00:37:03,680 --> 00:37:05,160
you're just going to
kind of duct tape it on.

876
00:37:05,160 --> 00:37:05,360
Right?

877
00:37:05,360 --> 00:37:07,770
Because that's going to give
you this big ball of mud code

878
00:37:07,770 --> 00:37:09,520
that's going to be
impossible to maintain.

879
00:37:09,520 --> 00:37:13,620
The idea with the refactoring is OK, I
know I only need, say, Twitter today,

880
00:37:13,620 --> 00:37:16,620
so I'm not going to do this
big abstraction that says,

881
00:37:16,620 --> 00:37:19,749
oh, let me have this abstraction layer
that will work with any social media

882
00:37:19,749 --> 00:37:22,290
network that I could ever possibly
think of it in the future,

883
00:37:22,290 --> 00:37:23,570
because that takes time.

884
00:37:23,570 --> 00:37:25,590
Let me just-- the simplest
thing that could possibly work

885
00:37:25,590 --> 00:37:26,710
is let me just make
it known with Twitter,

886
00:37:26,710 --> 00:37:28,293
because that's all I need to do today.

887
00:37:28,293 --> 00:37:33,680
Then tomorrow, we realize OK, we do
need to make this work with Facebook.

888
00:37:33,680 --> 00:37:37,530
So refactoring would say, let me revisit
the design before I even add Facebook,

889
00:37:37,530 --> 00:37:40,370
and say given that I
know that now I need

890
00:37:40,370 --> 00:37:43,590
to handle most multiple social networks,
what would the optimal design look

891
00:37:43,590 --> 00:37:44,090
like?

892
00:37:44,090 --> 00:37:46,620
Let me refactor the code
to handle that design,

893
00:37:46,620 --> 00:37:48,830
and then I can plug
Facebook functionality in.

894
00:37:48,830 --> 00:37:50,005
Does that make sense?

895
00:37:50,005 --> 00:37:52,860

896
00:37:52,860 --> 00:37:56,870
>> So a lot of people think, when they
hear something like emergent design,

897
00:37:56,870 --> 00:37:59,610
that you're doing less design
or that you're just hacking.

898
00:37:59,610 --> 00:38:01,850
But the truth is you're
actually doing more design.

899
00:38:01,850 --> 00:38:03,280
It's sort of the same
thing with planning, right?

900
00:38:03,280 --> 00:38:04,400
You're actually doing
more planning-- it's

901
00:38:04,400 --> 00:38:06,250
just that instead of
doing it all up front,

902
00:38:06,250 --> 00:38:10,470
you're doing it continuously
as you go along.

903
00:38:10,470 --> 00:38:16,150
>> So I think it's really great
that you guys are taking CS50,

904
00:38:16,150 --> 00:38:20,110
because I hear this so many times
a day, I can't even tell you.

905
00:38:20,110 --> 00:38:22,860
People come up to me and they say,
Abby, I've got this great idea!

906
00:38:22,860 --> 00:38:24,390
All I need is a developer.

907
00:38:24,390 --> 00:38:28,000
And I kind of want to shoot myself
in the head when I hear that.

908
00:38:28,000 --> 00:38:31,380
>> Because that kind of
assumes-- they'll come up,

909
00:38:31,380 --> 00:38:33,580
and they'll be like I have
the idea all figured out.

910
00:38:33,580 --> 00:38:34,705
I've got the business plan.

911
00:38:34,705 --> 00:38:35,540
I've got the design.

912
00:38:35,540 --> 00:38:37,440
I just need a developer to
go code it for me, right?

913
00:38:37,440 --> 00:38:40,148
And it's just assuming that they've
got all the answers up front,

914
00:38:40,148 --> 00:38:42,120
and this person can just
go code it for them,

915
00:38:42,120 --> 00:38:48,180
and they're going to make
a million dollars-- which

916
00:38:48,180 --> 00:38:50,850
just doesn't take into
fact all the uncertainties.

917
00:38:50,850 --> 00:38:54,060
>> So if we kind of look at the steps
of development-- and I apologize.

918
00:38:54,060 --> 00:38:55,480
This is a little waterfall-y.

919
00:38:55,480 --> 00:38:59,780
But what typically happens is you figure
out OK, this is what I want to code.

920
00:38:59,780 --> 00:39:02,380
You take some time to
develop it, test it.

921
00:39:02,380 --> 00:39:04,494
Quality assurance is testing it.

922
00:39:04,494 --> 00:39:06,660
And then once you've got
an entire release together,

923
00:39:06,660 --> 00:39:07,450
which might take a month.

924
00:39:07,450 --> 00:39:08,533
It makes two three months.

925
00:39:08,533 --> 00:39:11,260
Then you release that out, right?

926
00:39:11,260 --> 00:39:14,660
>> But if we say, OK, let's
think about how do we

927
00:39:14,660 --> 00:39:16,650
maximize the learning that happens here?

928
00:39:16,650 --> 00:39:20,770
Because if we just go heads-down for
three months or a year or something

929
00:39:20,770 --> 00:39:22,770
and put some code out
there and it doesn't work,

930
00:39:22,770 --> 00:39:24,720
then we're kind of screwed, right?

931
00:39:24,720 --> 00:39:27,044
So where does the
learning happen in here?

932
00:39:27,044 --> 00:39:28,960
Some learning happens
when we do requirements,

933
00:39:28,960 --> 00:39:32,210
because we're talking to customers, and
we're trying to understand about them.

934
00:39:32,210 --> 00:39:34,192
But the reality is that
most learning doesn't

935
00:39:34,192 --> 00:39:36,400
happen until we actually
put something in their hands

936
00:39:36,400 --> 00:39:38,330
and see how they use that.

937
00:39:38,330 --> 00:39:42,840
And so what this means is
that the time, the places

938
00:39:42,840 --> 00:39:47,550
that we spend the most time-- which is
development and QA or testing-- there's

939
00:39:47,550 --> 00:39:49,710
very little learning that happens.

940
00:39:49,710 --> 00:39:53,830
>> And so if we look at this and
say how can we maximize learning?

941
00:39:53,830 --> 00:39:57,580
Or how can we reduce the time
that happens between learning?

942
00:39:57,580 --> 00:40:00,340

943
00:40:00,340 --> 00:40:02,264
A great thing is continuous deployment.

944
00:40:02,264 --> 00:40:04,930
I don't know if you guys have
heard about continuous deployment.

945
00:40:04,930 --> 00:40:07,947
So the idea with that-- instead
of saying, OK, we're going to go.

946
00:40:07,947 --> 00:40:09,530
We have this releases at three months.

947
00:40:09,530 --> 00:40:11,120
We're going to build
all the features for it.

948
00:40:11,120 --> 00:40:12,600
And then only at the
end of the release are

949
00:40:12,600 --> 00:40:14,190
we are going actually
push that into production

950
00:40:14,190 --> 00:40:15,400
and put it in front of users.

951
00:40:15,400 --> 00:40:19,460
>> The idea with continuous deployment
is taking that to the other extreme.

952
00:40:19,460 --> 00:40:22,460
So are you guys familiar
with the version control?

953
00:40:22,460 --> 00:40:25,010
So ideally, when you're working
on your code, every time you

954
00:40:25,010 --> 00:40:27,230
add some new functionality, you're
gonna check it into version control.

955
00:40:27,230 --> 00:40:28,890
So if you screw something
up, you can always go back.

956
00:40:28,890 --> 00:40:31,440
Or you can see what changed,
if something's broken.

957
00:40:31,440 --> 00:40:33,335
>> So the idea with
continuous deployment is

958
00:40:33,335 --> 00:40:35,750
as soon as you check something
into version control,

959
00:40:35,750 --> 00:40:37,692
it pushes the code to a staging server.

960
00:40:37,692 --> 00:40:40,900
It's going to run automated tests on
it, make sure you didn't break anything.

961
00:40:40,900 --> 00:40:42,875
If you didn't break anything,
it's going to push it right out

962
00:40:42,875 --> 00:40:43,640
from the production.

963
00:40:43,640 --> 00:40:44,170
>> So boom.

964
00:40:44,170 --> 00:40:45,586
It's in the hands of the customer.

965
00:40:45,586 --> 00:40:48,020

966
00:40:48,020 --> 00:40:48,784
Very different.

967
00:40:48,784 --> 00:40:51,700
But if we do this, if we're pushing
things out to the customer as fast

968
00:40:51,700 --> 00:40:54,410
as possible, then we're getting
the code into their hands.

969
00:40:54,410 --> 00:40:56,470
We can see how they're
working with them,

970
00:40:56,470 --> 00:40:58,220
and we can really maximize learning.

971
00:40:58,220 --> 00:41:01,610
>> So I'm going to talk through
this a little bit more,

972
00:41:01,610 --> 00:41:05,162
because I don't know if that
was-- continuous deployment

973
00:41:05,162 --> 00:41:06,370
can be pretty extreme, right?

974
00:41:06,370 --> 00:41:08,380
And that can be pretty tough to do.

975
00:41:08,380 --> 00:41:12,217
So people, companies usually kind of
start with continuous integration,

976
00:41:12,217 --> 00:41:13,550
and they work their way forward.

977
00:41:13,550 --> 00:41:17,800
>> So continuous integration is this
concept that's kind of the first part

978
00:41:17,800 --> 00:41:18,900
that I talked about.

979
00:41:18,900 --> 00:41:21,550
So the idea with
continuous integration is

980
00:41:21,550 --> 00:41:23,192
you still have your release schedule.

981
00:41:23,192 --> 00:41:26,400
You're going to release every two weeks
or every three months or whatever is.

982
00:41:26,400 --> 00:41:28,740
>> But every single time
someone checks some code in,

983
00:41:28,740 --> 00:41:31,450
it does push the code
onto a staging server.

984
00:41:31,450 --> 00:41:34,170
The staging server looks
like production and it

985
00:41:34,170 --> 00:41:37,070
runs a series of automated tests
on them to make sure nothing broke.

986
00:41:37,070 --> 00:41:39,430
If something broke, then it's
going to let everybody know hey,

987
00:41:39,430 --> 00:41:40,330
the build was broken.

988
00:41:40,330 --> 00:41:42,970
And everybody has stop
and make sure it's fixed.

989
00:41:42,970 --> 00:41:48,750
>> So that way, you're always guaranteeing
that everything that you check in

990
00:41:48,750 --> 00:41:54,213
is keeping the code at an OK state.

991
00:41:54,213 --> 00:41:58,109
Then when you're ready to release it in
the fraction, you realize everything.

992
00:41:58,109 --> 00:42:00,900
Continuous delivery is sort of the
next step in this process, which

993
00:42:00,900 --> 00:42:06,210
is that every time you check-- it says
the same thing-- every time we check

994
00:42:06,210 --> 00:42:10,010
something into version control, it
pushes it to the staging server.

995
00:42:10,010 --> 00:42:11,386
It runs the tests on it.

996
00:42:11,386 --> 00:42:16,270
>> But the culture is set
as such that you always

997
00:42:16,270 --> 00:42:19,370
keep the code so that it can be
pushed to production at any time.

998
00:42:19,370 --> 00:42:21,440
So with continuous integration,
you might have a road map and say,

999
00:42:21,440 --> 00:42:23,800
we're only going to push it
to production in three months.

1000
00:42:23,800 --> 00:42:24,330
Right?

1001
00:42:24,330 --> 00:42:26,460
It doesn't really have to be
ready to be seen by a customer.

1002
00:42:26,460 --> 00:42:28,793
But with this, you're saying
at any given point in time,

1003
00:42:28,793 --> 00:42:31,020
you can be like yep, I'm
happy with this feature set,

1004
00:42:31,020 --> 00:42:32,140
even though we're only two weeks in.

1005
00:42:32,140 --> 00:42:34,390
I'm going to go ahead and
push it out to the customer,

1006
00:42:34,390 --> 00:42:35,820
and I know it's going to be OK.

1007
00:42:35,820 --> 00:42:40,800
>> And so you might have something
like switches in your code

1008
00:42:40,800 --> 00:42:42,780
that say for features
that are only half done.

1009
00:42:42,780 --> 00:42:43,710
They're not actually visible.

1010
00:42:43,710 --> 00:42:45,280
Why is it visible to the customer yet?

1011
00:42:45,280 --> 00:42:46,238
Or something like that.

1012
00:42:46,238 --> 00:42:48,450
But you always make sure
that you don't have anything

1013
00:42:48,450 --> 00:42:52,460
that's in this weird state, because it
can push out to production at any time.

1014
00:42:52,460 --> 00:42:58,310
>> And just once you're in, you've kind
of gotten everybody used to that idea

1015
00:42:58,310 --> 00:43:02,649
that you're always coding such that
it's ready to go out into production.

1016
00:43:02,649 --> 00:43:05,190
Then it's not so hard to move
to continuous deployment, which

1017
00:43:05,190 --> 00:43:07,231
is that every single time
you check something in,

1018
00:43:07,231 --> 00:43:09,560
as long as the test passed,
it goes out to production.

1019
00:43:09,560 --> 00:43:11,660
Does that kind of make sense?

1020
00:43:11,660 --> 00:43:14,490
>> So it can still be really
scary concept, but it's

1021
00:43:14,490 --> 00:43:16,790
interesting to look at how
some companies are doing it.

1022
00:43:16,790 --> 00:43:18,760
So Etsy does a really
good job with this.

1023
00:43:18,760 --> 00:43:21,270
If you're interested,
they've got a blog that

1024
00:43:21,270 --> 00:43:24,530
talks about how they do continuous
deployment, which is really awesome.

1025
00:43:24,530 --> 00:43:29,010
They deploy to production up
to 50 times a day-- right?

1026
00:43:29,010 --> 00:43:32,920
Which is crazy-- can you imagine if
you go to the Etsy website, 50 times

1027
00:43:32,920 --> 00:43:36,470
in day, that site is being
updated behind the scenes.

1028
00:43:36,470 --> 00:43:43,430
>> And in 2011, they deployed 10,000
times over the year with 100 engineers.

1029
00:43:43,430 --> 00:43:47,240
And what they said is contrary to
what you might think-- like oh my god,

1030
00:43:47,240 --> 00:43:47,920
that's terrible!

1031
00:43:47,920 --> 00:43:50,030
The code, the site's
going to be a disaster.

1032
00:43:50,030 --> 00:43:54,330
They said actually, once you're
deploying that often, the system is

1033
00:43:54,330 --> 00:43:57,610
so much more stable, they actually
call it confidence as a service.

1034
00:43:57,610 --> 00:44:02,930
Because when we deploy, we've
already done this 9,999 times.

1035
00:44:02,930 --> 00:44:05,670
We got this.

1036
00:44:05,670 --> 00:44:08,750
>> It also makes it so much easier
for them to experiment with things.

1037
00:44:08,750 --> 00:44:12,160
So what they said before is they
used to release to production

1038
00:44:12,160 --> 00:44:14,190
every two weeks or every month.

1039
00:44:14,190 --> 00:44:16,200
And you guys might
imagine if you've ever

1040
00:44:16,200 --> 00:44:18,350
got a deadline for a big
project you're working on,

1041
00:44:18,350 --> 00:44:20,990
and you have this list of things
that you want to get done,

1042
00:44:20,990 --> 00:44:22,320
and then as it gets
closer to the deadline,

1043
00:44:22,320 --> 00:44:23,900
the list starts shrinking a little bit.

1044
00:44:23,900 --> 00:44:25,340
Like well, maybe I don't
really need to do this.

1045
00:44:25,340 --> 00:44:26,470
Maybe I don't really need to do that.

1046
00:44:26,470 --> 00:44:27,590
>> So that's what they said would happen.

1047
00:44:27,590 --> 00:44:29,630
As they'd get closer to the
release-- and it was such a big deal.

1048
00:44:29,630 --> 00:44:30,990
They had to get the release out on time.

1049
00:44:30,990 --> 00:44:32,573
But they'd start paring away features.

1050
00:44:32,573 --> 00:44:35,230
And so they actually did less
features, because they were only

1051
00:44:35,230 --> 00:44:37,637
releasing every two weeks or a month.

1052
00:44:37,637 --> 00:44:39,345
Now that they're
releasing so many times,

1053
00:44:39,345 --> 00:44:42,137
it gives them this flexibility
to say, you know what?

1054
00:44:42,137 --> 00:44:43,970
We want to build a new
feature, but we don't

1055
00:44:43,970 --> 00:44:45,810
know if we should put
a lot of time into it.

1056
00:44:45,810 --> 00:44:47,940
Let's put out this really
minimum version of the feature

1057
00:44:47,940 --> 00:44:50,648
and see if anyone even clicks on
it, if anyone's even interested.

1058
00:44:50,648 --> 00:44:53,630
If they are, then we can either
pull it back and build it out,

1059
00:44:53,630 --> 00:44:56,250
or we can very quickly
add new features to it.

1060
00:44:56,250 --> 00:45:00,650
>> And so they said it just gave them so
much more flexibility to experiment.

1061
00:45:00,650 --> 00:45:04,990
And so it's really interesting to
see bigger companies doing that.

1062
00:45:04,990 --> 00:45:10,670
And at a startup, especially, where it's
so important to learn what's going on,

1063
00:45:10,670 --> 00:45:13,600
it can be really effective.

1064
00:45:13,600 --> 00:45:16,160
And then coming back
to our Kanban board.

1065
00:45:16,160 --> 00:45:16,939
>> It's interesting.

1066
00:45:16,939 --> 00:45:18,980
A lot of times, when people
do a board like this,

1067
00:45:18,980 --> 00:45:21,680
there's a lot of debate over
what the Done column means.

1068
00:45:21,680 --> 00:45:23,760
So OK, I'm working on a task.

1069
00:45:23,760 --> 00:45:25,317
Is it done when its code complete?

1070
00:45:25,317 --> 00:45:28,150
Is it done when someone's reviewed
it and it feels like it's tested?

1071
00:45:28,150 --> 00:45:30,010
Is it done when it goes
out into production?

1072
00:45:30,010 --> 00:45:32,370
>> And so a lot of startups
will say, you know what?

1073
00:45:32,370 --> 00:45:36,584
We're going to add a new column in
here, which is a learning column.

1074
00:45:36,584 --> 00:45:39,250
It's not actually done until we've
not only put into production,

1075
00:45:39,250 --> 00:45:41,070
we've put it in customers'
hands-- but we've actually

1076
00:45:41,070 --> 00:45:42,720
learned from how they've used it.

1077
00:45:42,720 --> 00:45:44,830
And what's really cool
about that is then,

1078
00:45:44,830 --> 00:45:47,415
we get to incorporate that
learning back into the cycle,

1079
00:45:47,415 --> 00:45:49,165
and say based on what
we've learned, based

1080
00:45:49,165 --> 00:45:53,680
on what we se-- how we see them use it--
we can figure out the next set to do.

1081
00:45:53,680 --> 00:45:56,970

1082
00:45:56,970 --> 00:46:02,980
>> So those are the patterns that I
have seen for successful innovation

1083
00:46:02,980 --> 00:46:06,380
across the startups that
have been successful.

1084
00:46:06,380 --> 00:46:09,835
I was going to also talk a
little bit about resources

1085
00:46:09,835 --> 00:46:12,873
that are available if you're
interested in doing a startup iLab.

1086
00:46:12,873 --> 00:46:17,590
But I can also stop it here, if you
guys have questions about what I talked.

1087
00:46:17,590 --> 00:46:18,350
Keep going?

1088
00:46:18,350 --> 00:46:20,093
OK.

1089
00:46:20,093 --> 00:46:21,450
>> [CHUCKLES]

1090
00:46:21,450 --> 00:46:24,320
>> OK, so do you know about the iLab?

1091
00:46:24,320 --> 00:46:24,870
OK, awesome.

1092
00:46:24,870 --> 00:46:28,345
So the iLab has awesome resources.

1093
00:46:28,345 --> 00:46:30,470
If you're looking to do a
startup, we have anything

1094
00:46:30,470 --> 00:46:31,880
from-- we do hacknights there.

1095
00:46:31,880 --> 00:46:33,755
Sometimes, we do
hackathons, if you just want

1096
00:46:33,755 --> 00:46:35,750
to go hack on cool projects with people.

1097
00:46:35,750 --> 00:46:36,724
>> We have workshops.

1098
00:46:36,724 --> 00:46:39,890
We have classes that re for credit that
are kind of cool on entrepreneurship

1099
00:46:39,890 --> 00:46:43,190
that are open to-- most of
those are open to everybody.

1100
00:46:43,190 --> 00:46:46,020
But we also have free workshops
a couple times a week,

1101
00:46:46,020 --> 00:46:48,040
that we just bring in
experts from the industry

1102
00:46:48,040 --> 00:46:52,750
to talk about anything-- from
technical concepts, to raising money,

1103
00:46:52,750 --> 00:46:54,830
to how to do sales.

1104
00:46:54,830 --> 00:46:57,340
>> Anything that you want
around startups, we

1105
00:46:57,340 --> 00:47:01,520
have experts and residents who
are available to do one-on-ones.

1106
00:47:01,520 --> 00:47:03,520
You can just sign up for
office hours with them.

1107
00:47:03,520 --> 00:47:05,103
You don't even have to have a startup.

1108
00:47:05,103 --> 00:47:08,140
Just if you've got ideas
and you want to balance--

1109
00:47:08,140 --> 00:47:11,150
get information or
insight from an expert

1110
00:47:11,150 --> 00:47:13,980
on the same thing-- sales, financing.

1111
00:47:13,980 --> 00:47:16,450
We get legal help.

1112
00:47:16,450 --> 00:47:19,830
You could sign up for those there.

1113
00:47:19,830 --> 00:47:21,370
We've always got stuff going on.

1114
00:47:21,370 --> 00:47:24,740
>> So if you're interested,
it's a really great resource.

1115
00:47:24,740 --> 00:47:27,846
You can go to our site.

1116
00:47:27,846 --> 00:47:29,220
The newsletter is really awesome.

1117
00:47:29,220 --> 00:47:31,770
I kind of usually hate
getting email, but it's cool.

1118
00:47:31,770 --> 00:47:34,279
We have so much going on, I
don't even know what all it is.

1119
00:47:34,279 --> 00:47:36,070
So if you sign up for
the newsletter, we'll

1120
00:47:36,070 --> 00:47:37,860
let you know every week what's going on.

1121
00:47:37,860 --> 00:47:43,830
You can also look at our calendar
to see what events are coming up.

1122
00:47:43,830 --> 00:47:47,710
>> And I am there to help if you
want to do a tech startup.

1123
00:47:47,710 --> 00:47:48,805
>> [CHUCKLES]

1124
00:47:48,805 --> 00:47:50,420
>> So that's what I've got.

1125
00:47:50,420 --> 00:47:53,720

1126
00:47:53,720 --> 00:47:55,220
>> [APPLAUSE]

1127
00:47:55,220 --> 00:47:56,720
>> [LAUGHS]

1128
00:47:56,720 --> 00:47:58,570
>> Thank you.

1129
00:47:58,570 --> 00:48:00,598