Tony Fadell tells builders not to surrender judgment to AI, arguing products should start from a real customer pain rather than easily generated output.
Fadell says the iPhone's physical keyboard debate at Apple was inconclusive in testing, so Steve Jobs made the call, a reminder that 1.0 products in new categories require opinion-based leadership.
Fadell blames Google's treatment of Nest as a stepchild for discontinuing the Nest smoke alarm without a successor, arguing home sensors are the context AI needs.
Tony Fadell on why AI cannot replace product judgment
Fadell argues that lasting products start from a real customer pain, rely on a small group with informed judgment, and fix the product before the business. AI makes output cheap, so judgment and craft become the scarce part. Builders who hand those over to machines ship things that are easy to copy and easy to discard.
Takeaways
Start from customer pain, then apply technology. Fadell says a product should begin with a real pain and use new technology to solve it. Nest began with energy bills. The thermostat cost 249 dollars and was claimed to save users 800 to 1,200 dollars a year.
Keep human judgment in charge when using AI. Fadell warns builders not to surrender cognitively to machines. A company built on easily generated output is throwaway. He calls the original Flighty luxury software because its craft and architecture are hard for AI to copy.
New categories force opinion-based decisions by few people. A 1.0 product has few analogs for data, so a small group must decide on informed judgment. Tests of virtual versus physical typing were inconclusive, so Steve Jobs made the iPhone keyboard call. Leaders should micromanage key decisions and their data, not daily operations.
Breakthrough products come from converging technologies. The iPod and iPhone rested on portable storage, MP3s, lithium-ion batteries, ARM processors, multi-touch and Wi-Fi. The first iPod sold mainly to Mac loyalists and needed three generations. Windows support and a 349 dollar price helped save Apple.
Fix the product first, then the business. Fadell says to make the product, fix it, then fix the business, with marketing designed in from the start. US marketing failed in Europe, and the 1,000 songs tagline came from a tiny, fast team. He faults OpenAI for adding product work late.
AI-written code creates technical debt. Fadell cites Dario Amodei's claim that Claude wrote 90 to 100 percent of Anthropic's code. He found engineers' reaction to the leaked Claude source brittle. He suggests using Claude Code for architecture and scoped tasks.
Nest's AI vision took fifteen years to arrive. Nest built AI in from 2010 but could not call its 2011 thermostat AI without alarming people. The smoke alarm was discontinued with no successor, which Fadell blames on Nest being a stepchild inside Google.
Where Fadell sees things heading
Voice becomes the primary input for future devices, while screens remain necessary.
Tools like Wispr Flow are great, but voice needs memory and intelligence behind it.
Fadell calls Humane's projection approach different from a screen, not better.
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Build Collective backs deep-technology companies such as Simbe Robotics and Greyparrot, and he invested early in Groq.
Designers should not addict users. The iPhone was not meant to be addictive, and Fadell says Google and Apple could do more to help people manage consumption.