Kevin Weil

AI product strategy requires a mindset shift because the underlying technology is a moving target rather than a fixed foundation.

"Everywhere I've ever worked before this, you kind of know what technology you're building on... but that's not true at all with AI. Every two months, computers can do something they've never been able..."

Adopt 'model maximalism' by building for the capabilities that are almost possible, as the model will likely catch up by launch.

"Our general mindset is in two months, there's going to be a better model and it's going to blow away whatever the current set of limitations are... If you're building and the product that you're build..."

The next phase of AI strategy involves combining powerful general models with proprietary, industry-specific data through fine-tuning.

"I think the future is really going to be incredibly smart, broad-based models that are fine-tuned and tailored with company-specific or use case-specific data so that they perform really well on compa..."

Complex AI products should be built as ensembles of specialized, fine-tuned models rather than a single generic model call.

"We use ensembles of models much more internally than people might think... If we have 10 different problems, we might solve them using 20 different model calls, some of which are using specialized fin..."

The reliability of a model's output (measured via evals) dictates the fundamental design and UX of the product.

"Writing evals is quickly becoming a core skill for product builders... you need to know whether your model is going to... get it right 60% of the time, you build a very different product than if the m..."

Use human interaction norms as a blueprint for designing AI user interfaces, especially for high-latency reasoning tasks.

"You can often reason about it the way you would reason about another human and it works... If you asked me something that I needed to think for 20 seconds to answer, what would I do? I wouldn't just g..."

Few-shot prompting with high-quality examples acts as a lightweight alternative to full model fine-tuning.

"You can do effectively poor man's fine-tuning by including examples in your prompt of the kinds of things that you might want and a good answer... the model really will listen and learn from that."

A new method of rapid prototyping where builders use AI to generate functional code through natural language and iterative feedback rather than manual coding.

"Instead of showing stuff in Figma, we should be showing prototypes that people are vibe coding over the course of 30 minutes to illustrate proofs of concept."

In fast-moving AI environments, high agency and comfort with ill-formed problems are more critical than traditional role definitions.

"High agency is something that we really look for, people that are not going to come in and wait for everyone else to allow them to do something, they're just going to see a problem and go do it... peo..."

Effective AI development requires breaking down the silos between research and product engineering to create a continuous feedback loop.

"The best products are going to be ENG product design and research working together as a single team to build novel things. So that's actually how we're trying to operate with basically anything that w..."

Market leadership in AI is driven by being the first to launch new capabilities and positioning the product as a universal tool.

"I think being first helps... we like being the first to launch new capabilities... ChatGPT can be this one-stop-shop where all the things that you want to do are possible."

Launching too many radical innovations simultaneously can trigger massive resistance; incremental introduction of change is often more effective.

"We tried to get it all at once. It was a new blockchain. It was a basket of currencies originally. It was integration into WhatsApp and Messenger... I think the whole world kind of went like, 'Oh my G..."

In AI, shipping early is essential because model capabilities and societal reactions can only be fully understood through public usage.

"We have this philosophy, we call iterative deployment, and the idea is we're all learning about these models together. So there's a real sense in which it's way better to ship something even when you..."