Lessons from Kevin Weil

On Product Philosophy and Strategy

“Wouldn't it be cool if…” is a terrible reason to build. Everything you build should be solving a problem for your customer. This principle, learned at Instagram, emphasizes that product development must be rooted in solving real user needs, not just chasing interesting ideas. [1]
Do the simple thing first. A core value at Instagram, this means boiling a problem down to its essential parts and getting that right before adding complexity. Systems naturally become more complex, so starting simple is crucial. [2]
Be unsentimental where it matters. Don't fall in love with something because you built it. Fall in love with it because it works for users. If a feature no longer serves its purpose, be prepared to remove it and build something better for customers. [3]
Product, done well, feels like magic. While the output can seem magical, the inputs are simple: a strong mission, good work, and consistency over time. [3]
You have to stay true to your core belief but be flexible on how you get there. Early-stage companies, in particular, should hold onto their founding mission while being open to significant pivots in their approach. [4]
The AI models you're using today are the worst AI models you will ever use for the rest of your life. This "model maximalism" philosophy at OpenAI acknowledges that while models aren't perfect, they are on a steep, continuous trajectory of improvement. [5]
If you're building something and you're afraid of the next model launch, you may not be building the right thing. Startups should focus on building products that will be enhanced by more capable future models, not those that simply patch the current model's weaknesses. [6]
Plans are useless, but planning is helpful. Weil subscribes to this Eisenhower quote, especially in the fast-changing world of AI. The value is in the process of stopping, assessing what worked and what didn't, and deciding what to do next, even if the plan itself will change. [7]
You build a very different product if a model gets it right 60% of the time versus 99.5% of the time. In the world of AI, product development must adapt to the "fuzziness" of inputs and outputs, unlike traditional software where systems are deterministic. [5]
It's better to make a decision, launch something, realize you got it wrong, and fix it, than to sit and talk in a circle for months. Weil learned the value of decisiveness and speed at Instagram, contrasting it with times he felt he was too indecisive at Twitter. [4]

On Leadership and Team Management

People are everything. No one actually does anything on their own. Everything you accomplish is with teams, and surrounding yourself with talented people who challenge you is key to personal and professional growth. [8]
As a leader, you need to ensure everyone knows the top 3 (or fewer) priorities. You should be able to ask anyone in the company what the most important things are, and they should know the answer. If they don't, you have more work to do as a leader. [3]
When scaling from one team to two, you have to fight the tendency to develop tribalism. As teams split, it's crucial to maintain open communication and keep everyone focused on the larger company mission to avoid an "us versus them" mentality. [4]
Have a regular, consistent forum for communication. Whether it's a weekly all-hands or a Monday note to the company, a consistent cadence for communication is vital to reinforce what's important and introduce change gradually. [4]
Hiring mistakes are situation-dependent; it doesn't mean the person isn't good. A person can be an amazing leader in one environment and not a good fit in another. It's about the match, not just the individual's talent. [4]
The best way to make AI safe is to get the models out there progressively. OpenAI's philosophy of iterative deployment involves co-evolving with society, learning from real-world use to discover capabilities and weaknesses together. [6][9]
Anyone can stop the line. To build an ethical product framework, every employee should feel empowered to call out an issue that needs review. [3]
A product role is all about people. This is a key insight from his time at Stanford's Entrepreneurial Thought Leaders program. [8]
Your job as a product manager is to be the voice of the customer. You need to understand what jobs your product is hired for and what problems it solves for them. [10]
Don't let your technical background short-circuit good ideas. Weil, coming from an engineering background, had to learn to stop dismissing ideas just because he knew they would be hard to implement. It's better to think about the problem you're solving first. [10]

On Career and Personal Growth

Go find a way to do what you wake up thinking about each day. You don't need to be a domain expert to start. If you believe in the mission and are extremely curious, you can do almost anything. [3]
Working in tech is a privilege and a responsibility. Realizing he could ship code and impact millions of people tomorrow, rather than waiting 40 years for a breakthrough in theoretical physics, was a pivotal moment for him. [3]
Seek variety in your career; you will learn new approaches at each stop. There is no single right way to do product, and experiencing different company cultures and methods is invaluable. [3]
I'm happiest if I feel like I half know what I'm doing and half have no idea. Being in a position where you are constantly learning and catching up is what drives hard work and happiness. [3]
Just keep running. Drawing a lesson from his passion for running 100-mile races, he notes that every person and company has ups and downs. The key is to not avoid them, but to choose to keep going. [3]
Every job I had was the biggest job I'd ever had, and it was intimidating. He overcame this feeling by talking to others and realizing that if people in even bigger roles could succeed, he could figure his out too. [2]
I'd rather hear from candidates about how they are going to make everyone have the stuff billionaires have instead of how they are going to eliminate billionaires. This tweet highlights a focus on creation and abundance over redistribution. [11]
My career has been completely unintentional. Weil's path from studying to be a math professor to leading product at top tech companies was not premeditated, but driven by seizing opportunities. [4]
Prototyping with AI models is an underused skill. For product managers, this can be a powerful way to quickly test and evaluate ideas. [10]
Writing effective evals (AI evaluation tests) is becoming a critical skill for product managers. In the AI era, being able to define and measure whether a model is successfully performing a task is fundamental to building good products. [5]

On Data, Users, and Feedback

Anecdotal feedback is valuable because it can get lost in the aggregate data. While data is essential, personal stories make you ask questions and develop hypotheses that lead to better products. [4]
When doing customer discovery, you can't go in with a hypothesis you are just trying to validate. Avoid asking leading questions. Your goal is to get underneath what customers say they want to understand the real problem, like needing to get somewhere faster, not just a "faster horse." [4]
I'm really tough to offend. I want feedback. After being head of product at Twitter, which he called the "world's best amplifier for people to tell us how bad we were at building product," he learned to embrace criticism. [4]
Testing is a way to understand what you're building; it's orthogonal to whether what you're testing is bold. You can test an ambitious feature on 99% of users or a small experiment in a test market. The goal is to get data to learn how it's working for people. [4]
The burden of product design is to make AI models work gracefully, even when they're not perfect. When models are only 60% right, human involvement is still needed, and the product design must account for that. [10]
Keep the purity of thinking about the problem you're trying to solve. Don't get bogged down early by whether a solution is easy or hard from an engineering perspective. Figure that part out later with the team. [10]
There's no science to knowing when to listen to users versus when to stick to your product strategy. It's a difficult balance, especially when users are resistant to change. [4]
Data tells you important things, and there's no excuse for not knowing it. While he values anecdotes, Weil emphasizes that a deep understanding of data is non-negotiable. [4]
The beginning of product development in AI can be uncertain due to the emergent properties of models. Sometimes you have to take a "wait-and-see" approach to understand a new model's full capabilities. [10]
You can often reason about how an AI product should work the way you would reason about another human. This has been a surprising and useful mental model for him at OpenAI. [5]

On the Future of AI and Technology

The crazy thing about OpenAI is that every two months, we make computers do something they have never been able to do before. The technological foundation is not fixed, which means you have to rethink your product constantly. [12]
AI-generated code could expand the pool of developers from 30 million to a billion, unlocking personalized software everywhere. This reflects a vision where AI democratizes the creation of software. [2]
It's our job that by the time other labs figure out our techniques, we're already three steps ahead. On competition in AI, he emphasizes the need to move fast and continuously innovate. [6]
The future of AI is going to be fun. He points to examples like AI acting as a real-time translator as just the beginning of how it will enhance human interaction and capabilities. [6]
I would be surprised if coding isn't fully automated before 2027. Given the rapid rate of progress, especially with reasoning models, he predicts an accelerated timeline for AI taking over most coding tasks. [12]
Opting out of AI will soon feel like refusing a smartphone. The technology will become so integrated into our lives that not using it will be a significant disadvantage. [2]
Bringing transparency is a massive positive, even if sometimes that means you capture some of the bad things that happen in the world. This was a key learning from his time at Planet, where satellite imagery reveals global events as they happen. [13]
More AI is better for the world. The philosophy at OpenAI is that getting AI into the hands of more developers will lead to more innovation that benefits everyone. [6]
Instincts can be helpful in product development, but only in about half of the job. This is especially true in AI, where the underlying technology is constantly and unpredictably evolving. [10]

Learn more:

Lessons from Kevin Weil

On Product Philosophy and Strategy

On Leadership and Team Management

On Career and Personal Growth

On Data, Users, and Feedback

On the Future of AI and Technology

Written by Antoine Buteau

Lessons from Bill Walsh

Lessons from Erik Torenberg