← 返回视频库

How to Build a Beloved AI Product - Granola CEO Chris Pedregal

中级 ⏱ 68:36

Granola is the rare AI startup that slipped into one of tech’s most crowded niches — meeting notes — and still managed to become the product founders and VCs rave about. In this episode, MAD Podcast host Matt Turck sits down with Granola co-founder & CEO Chris Pedregal to unpack how a two-person team in London turned a simple “second brain” idea into Silicon Valley’s favorite AI tool. Chris recounts a year in stealth onboarding users one by one, the 50 % feature-cut that unlocked simplicity, and why they refused to deploy a meeting bot or store audio even when investors said they were crazy. We go deep on the craft of building a beloved AI product: choosing meetings (not email) as the data wedge, designing calendar-triggered habit loops, and obsessing over privacy so users trust the tool enough to outsource memory. Chris opens the hood on Granola’s tech stack — real-time ASR from Deepgram & Assembly, echo cancellation on-device, and dynamic routing across OpenAI, Anthropic and Google models — and explains why transcription, not LLM tokens, is the biggest cost driver today. He also reveals how internal eval tooling lets the team swap models overnight without breaking the “Granola voice.” Looking ahead, Chris shares a roadmap that moves beyond notes toward a true “tool for thought”: cross-meeting insights in seconds, dynamic documents that update themselves, and eventually an AI coach that flags blind spots in your work. Whether you’re an engineer, designer, or founder figuring out your own AI strategy, this conversation is a masterclass in nailing product-market fit, trimming complexity, and future-proofing for the rapid advances still to come. Hit play, like, and subscribe if you’re ready to learn how to build AI products people can’t live without. Granola Website - https://www.granola.ai X/Twitter - https://x.com/meetgranola Chris Pedregal LinkedIn - https://www.linkedin.com/in/pedregal X/Twitter - https://x.com/cjpedregal FIRSTMARK Website - https://firstmark.com X/Twitter - https://twitter.com/FirstMarkCap Matt Turck (Managing Director) LinkedIn - https://www.linkedin.com/in/turck/ X/Twitter - https://twitter.com/mattturck LISTEN ON: Spotify - https://open.spotify.com/show/7yLATDSaFvgJG80ACcRJtq Apple - https://podcasts.apple.com/us/podcast/the-mad-podcast-with-matt-turck/id1686238724 00:00 - Introduction: The Granola Story 01:41 - Building a "Life-Changing" Product 04:31 - The "Second Brain" Vision 06:28 - Augmentation Philosophy (Engelbart), Tools That Shape Us 09:02 - Late to a Crowded Market: Why it Worked 13:43 - Two Product Founders, Zero ML PhDs 16:01 - London vs. SF: Building Outside the Valley 19:51 - One Year in Stealth: Learning Before Launch 22:40 - "Building For Us" & Finding First Users 25:41 - Key Design Choices: No Meeting Bot, No Stored Audio 29:24 - Simplicity is Hard: Cutting 50% of Features 32:54 - Intuition vs. Data in Making Product Decisions 36:25 - Continuous User Conversations: 4-6 Calls/Week 38:06 - Prioritizing the Future: Build for Tomorrow's Workflows 40:17 - Tech Stack Tour: Model Routing & Evals 42:29 - Context Windows, Costs & Inference Economics 45:03 - Audio Stack: Transcription, Noise Cancellation & Diarization Limits 48:27 - Guardrails & Citations: Building Trust in AI 50:00 - Growth Loops Without Virality Hacks 54:54 - Enterprise Compliance, Data Footprint & Liability Risk 57:07 - Retention & Habit Formation: The "500 Millisecond Window" 58:43 - Competing with OpenAI and Legacy Suites 1:01:27 - The Future: Deep Research Across Meetings & Roadmap 1:04:41 - Granola as Career Coach?

字幕文本（200 句）

AI is going to let humans work
0:02
differently, think differently. There
0:03
needs to be a tool that supports that
0:05
and that's what we want to build. So
0:06
this idea of a contextually aware
0:08
workspace like AI powered workspace like
0:10
that's what we wanted to build with
0:12
granola. And we said okay great perfect
0:14
we got the vision that's what we want to
0:15
build. Where the heck do we start?
0:17
>> Hi I'm Matt Turk from Firstark. Welcome
0:19
to the Matt podcast. My guest today is
0:21
Chris Pedreal the CEO of Granola. In
0:24
just over a year since he was launched,
0:26
Granola has emerged from the crowded
0:28
category of AI notetakers as a bit of a
0:31
darling in Silicon Valley. Not just as a
0:34
hut AI startup, but as an AI product
0:36
that many in tech circles use rapidly
0:39
every day and often describe as
0:41
life-changing. This episode is a
0:43
masterass in how to build a beloved
0:45
product in the age of AI, full of
0:47
practical tips and lessons that Chris
0:50
has learned along the way, including how
0:52
to achieve simplicity in product design.
0:55
>> We looked at it all and we cut out 50%
0:57
of it. We basically redesigned and cut
0:59
out 50%.
1:00
>> Knowing when to exit stealth
1:02
>> in this really, really busy market,
1:04
launching something more polished so
1:06
that when people use it, they're they're
1:08
wowed by it is is a way to stand out.
1:10
and what it's like being a let entrance
1:13
to a category and then having to compete
1:15
with a bunch of big companies including
1:16
potentially the open AIs of the world.
1:19
>> Open AI is going to try to do everything
1:22
to everyone. And I think the question is
1:24
can we do something way better for a
1:28
specific use case and a specific type of
1:30
user?
1:31
>> There's a lot to learn for AI builders
1:32
in this one. Please enjoy this great
1:34
conversation with Chris.
1:37
>> Hey Chris, welcome.
1:38
>> Thanks Matt. All right. So, not to
1:41
fanboy you from the very beginning of
1:43
this conversation, but I have to say I'm
1:45
a very rabid user of Granola and
1:48
actually our entire firm at First Mark
1:50
um is and um you know when I start when
1:53
I started using Granola a few months
1:54
ago, I thought I was pretty cool, a
1:56
pretty early adopter kind of kind of
1:58
situation. And then there was uh this
2:00
article in the information a couple
2:02
weeks ago that basically said well
2:04
everybody in Silicon Valley uses the
2:06
product all the time. So uh maybe not so
2:09
much of a of an early adopter from that
2:11
perspective after all. Just curious
2:13
about how that feels uh as a as a
2:15
founder to have a product that's just
2:18
widely embraced by our entire at least
2:21
little tech ecosystem.
2:22
>> It is it feels both amazing and and
2:24
daunting is the honest response. I uh we
2:27
did this this was last maybe November uh
2:32
um so we're based in London, right? And
2:34
we went to SF for a board meeting and
2:37
someone on the team said, "Hey, should
2:38
we rent out a bar and just email users
2:40
and say if you'd want to come and we'
2:43
like sure, you know, and we thought we
2:44
thought like five people would show up."
2:46
And like this two-fol bar was just full
2:48
of full of people. And um and then they
2:51
were just the the level of detail with
2:53
which they were talking about the
2:54
product or things we should change or
2:56
things that they had noticed. It it
2:58
really hit me cuz it when you you build
3:00
a product for people, but really being
3:03
in a room with that community all at
3:05
once, it it made me realize there's oh
3:08
there's something special that's
3:09
happening here that we didn't
3:10
necessarily design for. It's kind of
3:11
organically happening and now it's kind
3:13
of our job to
3:14
>> to follow that.
3:15
>> Yeah. and and one amazing which uh I'm
3:19
sure you've heard tons but it's just my
3:21
personal experience and looking online
3:23
and talking to people a description of
3:25
the granola experience that keeps coming
3:27
back is life-changing
3:29
uh which is insane but that's again
3:32
truly my experience I tweeted that at
3:34
some point I was uh you know all my life
3:36
a rabid uh notetaker that's how my brain
3:40
works and it helps me think through the
3:42
meeting or whatever I'm listening to uh
3:44
and pretty pretty much overnight that
3:46
lifelong habit just disappeared once I
3:49
tried granola a couple of times and
3:51
trusted it which u again not too fine
3:54
for you but uh it's it's been an
3:55
incredible experience.
3:56
>> I appreciate that. I think um I think
3:58
that will I think that speaks to the
4:00
moment in history that we're living in
4:02
where AI now has all these capabilities
4:06
um that are going to transform the way
4:08
we work and the way we think and
4:10
hopefully you that experience you just
4:12
described with granola will keep
4:14
happening in many more aspects of your
4:16
work life as you're you're basically
4:18
able to outsource lower level tasks and
4:21
allow you to think higher level um which
4:24
is why I think it's such an exciting
4:25
time to be to be building and to be
4:27
living quite frankly
4:28
>> I think when you released your uh team
4:31
product a few months ago you used the
4:33
expression second brain and it's it's
4:36
basically what it feels like and in some
4:39
ways a life-changing part does have a
4:41
little bit of a daunting aspect as a
4:43
user because it sort of feels like
4:45
you're outsourcing
4:47
your memory to technology and memory is
4:50
such a part of who we are as humans and
4:52
that's how you know humans survived for
4:55
for centuries like whoever was able to
4:57
remember facts, was was was able to just
5:00
function well in in society. So, it it
5:02
does feel like a like a wonderful tech
5:04
journey, but like something a little
5:06
more than that actually possibly.
5:07
>> I completely agree and I think there's
5:09
going to be as as uh as AI gets smarter
5:14
and we build more and more tools and
5:15
workflows on top of it, we're constantly
5:18
going to be letting go of things that we
5:20
used to do and letting machines do that
5:22
for us. And there are times when I think
5:25
that's ne incredibly beneficial and I
5:28
think there are probably times where
5:29
that's harmful, right? Like ne negative
5:32
beneficial. Uh the example I always use
5:35
is um Google maps, right, on on the
5:38
phone. So, it's very clear there were
5:39
cities I lived in before Google Maps
5:42
came out and I can go back to them. I
5:44
can still navigate them without a map
5:46
and there are the cities I've lived in
5:48
since Google Maps have come out and I
5:50
have like there's a very small area of
5:51
that city that I can navigate without
5:53
without a map. Um, so you could say, oh,
5:55
my my navigation skills have atrophied
5:58
100%. The number of hours I've walked
6:00
around lost in a city trying to find a
6:02
place has also gone down dramatically,
6:04
right? So it's like I'll I'll take in
6:06
that instance I'll definitely take that
6:07
trade-off. Um but I think you have to be
6:10
thoughtful case by case about what tools
6:11
you use and what you decide to
6:13
outsource.
6:14
>> And do we think long term that uh what
6:16
happens like we become we have more time
6:18
to actually think and reason but uh it
6:21
sort of feels like models are doing that
6:23
for us as well. I mean I I think this is
6:25
the this is why I'm so excited to be
6:28
building Grdola right now to be working
6:29
in the space because I I think that
6:31
future is kind of up to us right in ter
6:33
like there's this great quote which is
6:35
we shape our tools and thereafter our
6:37
tools shape us and when you think about
6:41
AI and um the future world and how it
6:44
fits into our society I I think there's
6:46
this big question of like what you know
6:47
what do we outsource to AI where does AI
6:50
replace humans and and where does it
6:51
augment humans and um where like me
6:55
personally, Sam Granola, we're really
6:57
big fans of the augmentation idea. This
6:59
goes back to Douglas Anglebart in the
7:01
50s, right? Augmenting human
7:02
intelligence. And his his view,
7:05
honestly, his stuff, I feel like people
7:06
don't talk about him enough. It's just
7:08
so inspiring. Like, he's known for being

在 Linglish 上深度学习这个视频

AI 助手 · 单词高亮 · 影子跟读 · 间隔重复 · 完全免费

立即免费学习