What product intelligence becomes
Three layers. Voice, behaviour, environment. Layer 01 is live. Here's where product intelligence goes from here.
At WPP in Singapore, I ran a product that tracked cultural trends.
Not through surveys or focus groups — through a network of scouts. People embedded in cities around the world, reporting from the frontlines of culture. What was emerging in Tokyo that hadn't reached New York yet. What was fading in London that was just arriving in São Paulo. The signal was ambient, distributed, accumulating continuously across hundreds of sources. No single data point meant much. The pattern was everything.
At the same time, I was leading large-scale sentiment and buzz tracking. What people were saying about brands, products, ideas — across media, across markets, across languages. The tools were good for the time. Nothing like what's possible today. But the obsession was the same: find the signal before it becomes obvious. See around the corner before everyone else does.
Years later, at TradeGecko, I'd sit on weekends and refresh the Tableau dashboard. Watching how leads were performing, how conversion moved, where the segments that might indicate PMF were hiding. I worked with analysts to build elaborate dashboards — combining event data with the feedback we were hearing, the patterns we were seeing in usage. Trying to reconcile two things that are sometimes the same and sometimes completely different: what people say, and what people do.
That gap — between voice and behaviour, between explicit feedback and revealed truth — is one of the most important signals a product team can have. And it's the one that most product intelligence tools still don't reach.
Circuit reaches it. Here's how.
Where we are now
February built the missing layer. March made it work for teams.
Today, Circuit is live and available for builders everywhere. Feedback flows in from every channel. Smart priorities surface what actually matters — ranked by signal strength, not volume or whoever spoke loudest. Specs generate grounded in the codebase. Customers hear back when what they asked for ships. And with every cycle, Circuit's Instinct layer learns — getting sharper at understanding what matters to this product, this team, these customers.
That's the loop. Running now.
But it's the beginning of what product intelligence becomes. Not the destination.
The three layers
Product intelligence isn't a single capability. It's a stack — three layers that compound on each other, each one surfacing a different kind of truth.
Layer 01 — What customers tell you
Live now.
The explicit voice. Feedback from every channel — scored, ranked, turned into codebase-grounded specs. Customer signal and product judgment, working together in every decision.
This is the foundation. It's what I spent 19 years doing manually with spreadsheets and exports and colour-coded tabs. Circuit does it automatically, at scale, continuously. The scout network I ran at WPP required hundreds of people. Circuit requires a one-line embed and a GitHub connection.
Most teams stop here — not because they want to, but because reaching further requires infrastructure that hasn't existed. Until now.
Layer 02 — What customers show you
Coming next.
Not what they say. What they do.
Usage patterns. Behavioural signals. The gap between what people praise in feedback and what they actually use in the product. A feature that customers love in surveys but never open. A workflow that nobody mentions but everyone depends on. The workaround that half your users have built and nobody has reported because they assume it's intentional.
I spent weekends refreshing dashboards trying to find these signals manually. Combining event data with feedback, looking for the places where voice and behaviour diverged — because that divergence is where the most important product decisions live. When what customers say and what customers do point in different directions, something true is being revealed.
Circuit will surface it automatically. Behaviour and voice, synthesised together. The gap made visible.
Layer 03 — What the environment reveals
On the horizon.
The Slack thread where three customers describe the same workaround without realising it's a product problem. The support conversation with a signal buried inside it that never makes it to the product team. The pattern accumulating in silence across channels nobody's monitoring — not because it's unimportant, but because no one has the bandwidth to listen everywhere at once.
At WPP, I had a network of human scouts doing this for culture. People embedded in the world, watching for patterns before they had names. It was powerful. It was also expensive, slow and impossible to scale.
Ambient product intelligence is the automated version of that network. Continuous, passive, always accumulating. Not waiting for feedback to be submitted — listening to everything, finding the patterns that haven't been reported yet, surfacing the problems before they become urgent.
Most tools show you the past. Layer 03 is built to know what's happening now — and what to build because of it.
Why this matters
The progression from Layer 01 to Layer 03 isn't just a product roadmap. It's a shift in what product intelligence means.
Layer 01 answers: what are customers telling us? Layer 02 answers: what are customers showing us? Layer 03 answers: what is the environment revealing that nobody has noticed yet?
Each layer surfaces a different kind of truth. Together they give a product team something that has never existed as infrastructure: a complete picture of what to build next — not based on who spoke loudest, not based on which dashboard someone refreshed on a weekend, but based on everything customers are saying, doing and revealing across every channel, continuously, automatically.
I built the manual version of this for 20 years. Spreadsheets. Dashboards. Scout networks. Sentiment tracking. Always trying to see the full picture with tools that only showed part of it.
Circuit is the infrastructure for the full picture.
That's what autonomous product intelligence means. Not a dashboard you open. A system that runs — ingesting signal, ranking priorities, generating specs, closing the loop — whether or not anyone is logged in.
The loop is running. The layers are building. That's the direction.
What is autonomous product intelligence? The full breakdown →
Catherine Williams-Treloar is the founder of Circuit — the AI product system that turns customer feedback into scored priorities and build-ready specs for Cursor and Claude Code. She has 20+ years leading product, insights, strategy and GTM at scale-ups and enterprises. Circuit was founded in Sydney in November 2025 and launched in February 2026.
Circuit turns customer feedback into ranked priorities and build-ready specs.
Get started free