We want the people who desire the outcome of a great startup:
an incredible product that solves problems, something that creates a durable and meaningful business
Introducing Meta Segment Anything Model 2 (SAM 2) — the first unified model for real-time, promptable object segmentation in images & videos.
SAM 2 is available today under Apache 2.0 so that anyone can use it to build their own experiences
Details ➡️ https://t.co/eTTDpxI60h
Execution is everything.
You can read all the business books and listen to all the podcasts but the real learning happens when you start building.
Execution isn't just a phase.
It's everything.
Startup ideas are worth zero without it.
What if we could universally recombine, insert, delete, or invert any two pieces of DNA?
In back-to-back @Nature papers, we report the discovery of bridge RNAs and 3 atomic structures of the first natural RNA-guided recombinase - a new mechanism for programmable genome design
I hate that the startup world puts investors on a pedestal. As a startup founder, it distracted me from serving my users. As an investor, it distracts my companies from serving their users. Investors aren’t the heroes of the startup world.
I’ve noticed a pattern with some AI startups that are producing flashy demos with very low customer engagement.
They are usually solving occasional problems.
An example is “use natural language to query a database”.
Most employees at a tech company don’t actually write SQL queries every day. They might want to write a query once a month and struggle with the syntax. This is an occasional problem.
Data scientists, on the other hand, write SQL queries every day, but the solution they’d want is a power tool for data scientists. It would probably have version control, performance monitoring and collaboration features. Not a chatGPT style interface for hobbyists and beginners.
LLMs are really shining right now where they are applied to repetitive work that’s done by teams of human for 8+ hours a day.
@greenliteai is automating KYB for banks
@SolveIntel is automating the production of legal patents
@markprompt and @Senso_AI are making customer service agents more effective.
People often ask what my #1 lesson from YC was.
I learned what speed looks like. It all came down to one moment.
It was just after group office hours and my cofounders and I were telling @Gustaf our plan:
“We’ll build self-serve, and then a few more table-stakes features, make the app more reliable, launch to YC in 2 weeks, then launch on product hunt in 3-4 weeks”
Gustaf: “You should launch to YC immediately, and Product Hunt Thursday”
We felt terrified.
We spouted excuses - the app wouldn’t deliver, people wouldn’t retain.
Gustaf listened, then said simply: more than anything else, you need to know whether people want what you’re building.
He was right. Our first launch - internal to YC - wasn’t great.
But we learned, refined our positioning, and our Product Hunt launch went viral / led to a monster seed raise from a16z…YC, and Gustaf particularly, felt like a friendly hand on your back pushing you uncomfortably fast.
Thank you Gustaf.
Everyone needs that person to push them to move faster.
Corollary:
You must rely on customers to ITERATE
but you cannot rely on customers to INNOVATE.
Customers are great to tell you how improve something that exists, but you better be the one who brings something new into existence. or you're dead.
You're trying to build a product the market wants, so that means we should ask the market what to build, right?
WRONG.
Building by survey is a sure path to a consensus product. By definition.
What's the alternative?
Start with an insight - an earned secret - then SHOW the market and see if it strikes a nerve.
"No customer ever asked Amazon to create the Prime membership program, but it sure turns out they wanted it, and I could give you many such examples." - Jeff Bezos.
It's the innovator's job to innovate, not the customer's
One my best programming learning experiences was git core development.
Making useful changes to git meant rapidly gaining deep understanding of how an often-inscrutable system works. Even found & fixed a security vulnerability!
In general, programming rewards learning by doing.
i love openai, and everything i’ve done over the past few days has been in service of keeping this team and its mission together. when i decided to join msft on sun evening, it was clear that was the best path for me and the team. with the new board and w satya’s support, i’m looking forward to returning to openai, and building on our strong partnership with msft.