20VC Newsletter - 25th August 2024
Here are the transcripts and top takeaways from 20VC episodes released this last week!
Monday’s episode with Aidan Gomez, CEO & Co-Founder @ Cohere.
Download the full transcript:
My 8 key takeaways:
It Is Total BS to Think AI Is Plateauing
There is lots of untapped potential in reasoners and planners.
We have only begun to focus on it in the past year.
There will be a massive change in capability when it comes out.
Is Compute the Answer to Performance
It is the most trustworthy, low risk, and dumbest strategy.
It is a compelling method for companies with deep pockets.
BUT it is extremely inefficient & there are better ways to scale.
Will There Be a World of Many Smaller Models or Fewer Larger Models
We are seeing a pattern where people prefer to prototype general models.
They do not like to fine-tune models for specific tasks.
There will likely be a mix of horizontal and vertical models.
It is Getting Harder to Deliver Gains in Models
Models used to be dumb & anyone off the street could spot errors & train it.
As they improved we needed smarter & smarter domain experts.
We’ll need scarcer pools of talent to teach the models incremental knowledge & it will be higher friction & more expensive.
Consolidation of Startup Model Providers is Coming
We are seeing a trend already: Adept to Amazon, Inflection to Microsoft…
It is really dangerous when you make yourself a subsidiary of your cloud provider.
It’s not a good business: Raising money from cloud providers is very different vs outside investors.
OpenAI Has Shelved AGI and is a Consumer Product Company
ChatGPT is a household name & they are killing it in consumer products..
This is necessary for the capital required to build AGI.
BUT it seems like they are prioritizing consumer products now & AGI has taken the back seat.
We Are Not in the Experimental Budget Phase Anymore for Enterprises
Last year was a proof of concept for everyone.
Now there is an urgency to get the tech into production.
Lots of enterprises are scared of being caught flat-footed.
Where is Value Today and Where Will It Be in the Long Term
There is value in the chip layer from all the demand.
We are also seeing value in application layers like ChatGPT’s subscription model.
Model layer is a good business long term but it is a low-margin & commoditized in the short term.
Wednesday’s episode with Aman Narang, CEO & Co-Founder @ Toast.
Download the full transcript:
My 7 key takeaways:
Why This $14BN Company Would Not Exist Without Speed
Too many people worry about getting the product right.
One of the most important things you can do is get something out & learn from the user.
We wouldn’t have existed today if we didn’t move fast & iterate.
The Man Who Bought 12% of a $14BN Company for $500K
We were trying to raise in 2012-2013
We were confident but almost everyone said no.
Steve Papa @ Endeca wrote us a $500K check at $3M.
The $14BN Company That Did Not Have a Good Business Model to Start
We realized within the first 10 customers that the financial model wasn’t working.
We were making $200 - $300 a month but there were too many customer service calls.
It wasn’t scalable at all.
Biggest Mistake: Not Hiring Managers Faster
There is also leverage in hiring managers vs. salespeople or engineers.
You need to move fast once you see signs of PMF.
Building out a team around us would have helped us move faster.
The Importance of Market Timing
3 Things:
Android
Cloud
Embedded payments
Without those 3 Toast wouldn’t have existed.
You Have to Sequence Strategy
We see many ways to expand the Toast platform across other geos & segments.
We need to be thoughtful about how to sequence this.
This is a long game: You underestimate what you can do in a decade and overestimate what you can do in a year.
This Public Company Creates Smaller Companies Within It
We have a New Ventures program for entrepreneurs.
We let them build companies as part of Toast but operate independently.
We want them to have the benefits of being part of Toast without the challenges of being part of a larger org.
Friday’s episode with Chad Peets, Managing Partner @ RPT Capital.
Download the full transcript:
My 6 key takeaways:
Why You Have to Hire a CRO Pre-Product
Founders do not know which product the market wants.
They cannot be the person doing customer discovery.
You need a sales expert to go through thousands of calls & evolve with the product.
Why Hiring Panels are BS
The decision should be binary: you either want to hire them or you don’t.
Anyone not uniquely qualified to qualify or sell should not be a part of the interview process.
Having a sales engineer on the panel is risky & slows everything down.
Why You Always Want to Start Mid-Market
You want to nail a small set of accounts with less demand.
You will learn what it takes to make enterprise customers happy.
This buys you time to build the product to go enterprise at a later date.
Any Inside Sales Rep Who Wants Remote is a Loser
They should recognize that by being in office, they will improve faster.
A daily 30-minute commute could significantly further their career.
If they do not want to make that sacrifice, I will not hire them.
Culture is BS and People Do Not Want to Work
Culture is not about the benefits & the work-life balance.
It is about meritocracy & being surrounded by world-class talent.
People are not willing to sacrifice for exceptional things anymore.
Biggest Advice to New Sales Leader
Take the time to invest in your career.
Reach out to people and learn as fast as you can.
You will become better than everybody else around you.
Let us know what your big takeaways from this week’s shows were in the comments below!
Thank you for reading, and don’t miss the great guests we have next week:
Imran Khan, Founder & CIO @ Proem
Arvind Narayanan, Director @ Princeton
Scott Gorlick, Fmr Head of Strategic Projects @ Uber
Thank you for reading 20VC.
This post is public so feel free to share it.
Thanks for reading 20VC! Subscribe for free to receive new posts and support my work.