Behind the build: Teresa Torres’ first AI product

Hey community 👋

In our latest webinar, we sat down with Teresa Torres, leading voice in product discovery and author of Product Discovery Habits, to hear the story behind her first AI product, Product Talk’s AI Interview Coach.

Her key lesson? Building with AI isn't just a new feature; it’s a fundamental shift in how we architect, synthesize, and define quality.

Here are my four favorite quotes from the session, and what they mean for product managers. 

1. “If I had to write my book, Continuous Discovery Habits, today, I would add a chapter on AI evals. They’re a new discovery habit.” 

Many teams accept an 80% quality output, but if AI is your differentiator, "good enough" is a failure.

• The new habit: AI Evals are now a mandatory discovery practice.
• The goal: Use evals as guardrails so a model never shows a customer a response that doesn't meet your criteria.

2. “We have to understand how to construct input for high-quality output, understand that the context window is limited, understand the concept of context rot, and how to manage it.”

Don’t overload your LLM with "mega-prompts." It leads to context rot and confusion.

• The fix: Deconstruct big tasks into a series of smaller, orchestrated LLM calls.

• Example: Teresa’s coach failed as one prompt; she had to break it into four separate calls to keep the AI focused.

3. “I want to get the most out of every interview because to me it's gold. So I'm going to mine it.”

"Lazy AI synthesis" is the most common anti-pattern today. If you spend time talking to a customer, don't outsource 100% of the thinking to a machine.

• The winning workflow: Do the synthesis yourself first, then ask the AI to do it, and compare the gaps. Use AI as "another perspective" to mine every bit of gold from an interview.

4. “Everybody needs to invest in their builder mindset. This means we have to get a little more comfortable with these technical building blocks.”

For Teresa, the "product PM vs. technical PM" debate is over. Every product manager needs their builder toolkit:

• Understand the "plumbing": How models work, how to manage context windows, and how to design APIs.

• Clarity is the ultimate skill: AI requires PMs to be "intent architects" who write exceptionally strong specs.

How are you balancing the speed of AI synthesis with the need to actually know your customer? How are you keeping the "human in the loop" without losing the efficiency gains? Share your thoughts in the comments.

Want to watch the full session with Teresa Torres? We’ve saved a recording for you here. 😊