When Not to Use AI: Product Judgment in the Age of AI Hype

• Not every problem wants a neural net.
• Not every feature needs a model.

In a world racing to bolt AI onto everything, the most underrated product skill right now is knowing when to say no.

Featuring an incredible panel of Founders/CEOs/CTOs:

• Nils Alstad — Founder / CEO of a stealth defense robotics startup; formerly CPO at Phantom Auto and customer ops lead at Canvas Robotics (acquired by Amazon)
• Fernando Nobre — CTO at Durable; formerly Applied Science Manager at Amazon and perception scientist at Canvas Robotics (acquired by Amazon)
• Lucas Thelosen — Founder / CEO at Gravity; formerly Head of Product, Data & AI at Google Cloud

This panel is for builders who’ve asked themselves:

• “Is this actually better… or just more complex?”
• “What happens when 99% accuracy fails on the 1% that matters?”
• “Are we solving a real user problem or just shipping AI theater?”

Our speakers have lived at the sharp edge of autonomy, robotics, data, and AI platforms where mistakes are expensive, safety matters, and predictability often beats cleverness. They’ve seen firsthand when deterministic systems outperform probabilistic ones, when rules beat learning, and when restraint is the real innovation.

### What we’ll dig into:

• When not to use AI and why that decision can save your product
• How to evaluate AI features using first-principles thinking
• Real examples where “simpler” won and “smarter” failed
• How to spot AI that’s solving problems vs. AI that’s just checking a box

If your backlog is full of AI ideas, your leadership is asking “Where’s our AI story?”, or you’re quietly worried about adding risk, inconsistency, or fragility to your product… this conversation is for you.