Build what your users actually need.
Because the wrong feature ships just as fast as the right one.
“Like getting a direct line into our users’ heads, and our revenue showed it: up 20%.”
The shift
Building got cheap. Deciding didn’t.
Coding agents ship whatever you describe, in minutes. The bottleneck moved upstream: not whether you can build it, but whether it’s the right thing to build. Most teams still decide on gut, copy a competitor, or ship and hope.
Diagnose
Dipio interviews the users who churn, stall, or switch, then reads the transcripts through behavioural science, so you get the real reason, not the surface complaint.
The real why, in their words
Decide
It turns the diagnosis into one ranked recommendation: the single change most likely to move behaviour, with the quotes and numbers that back it.
A prioritised, evidence-backed call
Build
It writes an agent-ready spec, hands it to your coding agents, and traces every change back to the evidence. A human signs off before anything ships.
A spec your agents build
How it works
One decision, start to finish.
A real one: a checkout problem, diagnosed from how users behave, then reviewed and signed off by your team before a line gets built. Every claim traces back to the evidence it came from.
Why
Evidence
What
Diagnosis
How
Intervention
Measure
Success Metrics
Caveat
Caution
Approved by PM and UXR
🦊🦉Approved by 3 engineers
🐼🦁🐨Ready for every major agent harness
Claude Code
Cursor
Conductor
OpenCodeOnce signed off, engineers pull the recommended interventions as agent-ready specs, or the evidence behind them, straight from their agent harness through the Dipio MCP. Nothing unapproved is ever exposed.
Dipio Evidence Loop
Every answer makes Dipio sharper on your users.
Real behaviour becomes evidence, evidence becomes decisions, and your agents ship them. And loop after loop, that evidence grows into an accurate simulation of your user base, grounded in how your users make decisions in real life, so you can test a change before you build it.
An AI interviewer holds real voice conversations with your users, far closer to ground truth than a self-reported survey.
The diagnosis pinpoints what blocks your users and what to build next, grounded in psychological and behavioural science.
The diagnosis becomes an implementation-ready spec, grounded in traceable evidence and ready for your coding agents.
Your coding agents and engineers build straight from the spec, with no translation step in between.
Every release ships as a measurable experiment, and the results become the next question to study.
Dipio Evidence Gate
Your agents build from proof, not vibes.
The approved decision drops straight into your coding agents. They build only what the evidence supports, trace every change back to it, and refuse when the research runs out. A human signs off before anything ships.
> add a “remind me later” button to the checkout flow
⏺Grounding this in your research first.
⏺dipio - research_context(question: “why users drop at checkout”)
⏺Evidence points to a price-preview step, not a reminder. Pulling the spec.
⏺dipio - get_spec(study: “checkout-abandonment”)
⏺Ran /speckit.specify with the Dipio spec → plan → tasks → implement
> also add an exit-intent discount popup
⏺dipio - research_context(question: “exit-intent discount”)
⏺I won’t invent it. Want me to draft a study and hand it to product to test this first?
Connect from Settings → Agents after signup.
Who it’s for
Product’s not the only team guessing.
Product, marketing, growth, and research run the same loop: turn how customers actually behave into the decision that comes next.
Product
Decide what to build next
Turn real user behaviour into a ranked view of what to build, and go from open question to shipped feature in one cycle, not one quarter.
Marketing
Know what actually moves customers
Ground campaigns and messaging in why customers really decide, not what a survey says they’d click.
Growth
Fix why users churn or never activate
Diagnose the behaviour behind the funnel, so you change the reason people drop, not the symptom the dashboard shows.
Research & strategy
Know why customers choose you, or don’t
Reconstruct the real decision: why people pick you, leave, or switch to a competitor, read through behavioural science.
Research
We introduced synthetic twins. Now we’re matching minds, not just answers.
Dipio introduced synthetic twins in 2024, and the method is peer-reviewed: our first paper is in press at Behavioural Public Policy (Cambridge University Press), grounded in behavioural science from the UK’s #1 ranked university in Psychology, the London School of Economics and Political Science.
Paper I
In pressCan AI represent the way real people think and decide?
We compared Dipio’s synthetic twins against 373 real people, across three products and four AI models. In aggregate, their answers tracked real humans closely. But the paper shows the limit too: matching what people say is not the same as matching how they think, and closing that gap is the work we’re on now.
From matching outputs to matching minds.
The field judges a synthetic user by one thin test: does its answer resemble a human’s? That says nothing about why it answered, or whether it would answer the same way tomorrow. We’re building the layer underneath: twins that reason from the same psychological mechanisms as real people, their traits, values, heuristics, and biases, so the resemblance holds when the question, the model, or the task changes.
Matching outputs is a benchmark. Matching minds is the science.
Twins Studio, the product built on this research, is in early access. It runs the interview step at simulation speed, so evidence arrives in minutes, not weeks.
Get started