Praxis Technology
The engine behind the diagnosis
Praxis is a proprietary behavioral-science diagnostic engine. It takes a real-world behavior problem, identifies the mechanism most likely driving it, and guides the work from diagnosis to a measured result. Client data never leaves the consultant's machine.
Every project starts with one question
Is it already running, or not launched yet? The answer sets the track: in operation or in design. Each has its own steps.
In operation
It is running, or it already happened.
In design
It has not launched yet.
Intake
The situation in your own words, plus a fixed set of details, completed together on an intake call. One format for every client.
Diagnose
Praxis names The Why and The How and places the case on the Praxis Matrix. The result is a hypothesis, with no confidence rating yet.
Validate
Praxis specifies the data to collect, and your real data is tested with established statistics. This is the only point where confidence is given.
Intervention
Drawn from solutions that researchers proposed for that mechanism. Nothing is proposed until your data supports the hypothesis.
Trial Design & Evaluation
The intervention is trialled, ideally against a control group, and measured over a set period. Did it work, and by how much?
Scaling
What worked is rolled out more widely.
Intake
The same intake, in the same format.
Gap Analysis
The policy or process is checked for behavioral gaps before it reaches people: which risks are covered, partly covered, or missing.
It is read through the same The Why and The How families as a running case, looking for the moments where people are likely to stall or drift:
- Steps that are unclear or take more effort than they should
- Timing, reminders, and defaults that work against the intended action
- Incentives, social cues, or trust issues that pull the other way
Each gap is named and explained, so the intervention has a clear target.
Intervention
Designed to close the gaps found, drawn from solutions that researchers proposed.
Trial Design & Evaluation
The first real test of the design, ideally against a control group, measured over a set period.
Scaling
What worked is rolled out.
Two answers: The Why and The How
Every diagnosis gives two answers. The Why is the theory that explains the behavior. The How is the practical framework you can act on. Either can come back as "none", and Praxis says so instead of forcing a fit.
Theory families
- Leadership and relationshipsHow people are led, and how they get on with those who lead them
- Motivation and goalsWhat people want, what they are asked to aim for, and whether the effort feels worth it
- Fairness and strainWhether decisions feel fair, and whether the work asks more than it gives
- Offer and outcomeWhether what people got matched what they expected
Practical framework families
- Context and choice designHow options are presented, timed and framed
- Process and accessThe steps, effort and obstacles between people and the action
- Offer and outcomeWhether the offer itself does what people need it to do
- None fitsPraxis says so instead of forcing one
Independent evidence only
Peer-reviewed research and non-partisan public-policy bodies. Advocacy think tanks are excluded, so no political position is built in.
The Praxis Matrix
Every case is placed where its Why meets its How: the family of theory that explains the behavior, and the family of frameworks that can change it.
Illustrative. Every case lands where its Why meets its How. When Praxis finds one without the other, it says so.
Honest by design
Most tools in this space overclaim. Praxis shows exactly what its conclusions rest on, and how strong the evidence is.
Hypothesis first
Diagnosis results are shown as unverified hypotheses, with no confidence rating. Confidence only comes from evidence: your data at Validate, or the trial for work still in design.
Evidence is graded
Every intervention is labelled by how much research stands behind it. It shows how well studied an idea is, not a promise that it will work.
Thin data is called thin
When your data is too small for a reliable result, Praxis says so, and reports a range instead of a falsely precise number.
Reasoning is always shown
Every diagnosis shows why it was reached, and each result can be confirmed or flagged by the consultant.
What the AI does, and doesn't
It does
- Make judgment calls: which Why, which How
- Show the reasoning behind every diagnosis
- Say "none" when no Why or How fits
It never
- Produces the numbers. Every figure comes from built-in statistical software
- Invents interventions. They come from a library of solutions researchers proposed
- Claims more certainty than the data supports
Where Praxis fits
Established behavioral-science units do rigorous, bespoke work, usually over weeks or months, built around randomized trials. Their strength is the trial. The first diagnosis is still expert judgment, redone for every project. Praxis adds a fast, standard front end to that work.
Praxis is
- The fast, standardized front end of behavioral work: ranked hypotheses in minutes, a tested diagnosis in about two weeks
- Every framework weighed equally, not just the cheapest to trial
- The Why and The How identified before any testing begins
- Its own trial loop for organizations that can't commission a large trial
Praxis is not
- A replacement for a full-scale randomized trial run by a major evaluation body
- A source of guaranteed outcomes
Privacy by design
No names. No identities. No cloud.
Runs on our machine, not the cloud
Praxis runs on the consultant's own machine. Client material is never sent to a third-party server.
Works only with your codes
Data is pseudonymous: Praxis sees only the codes you assign, never names or identities. The key that links the two stays with you.
You stay in control
Personal data never reaches Praxis. You decide what is shared, and you carry out any intervention yourselves.
See it on your own problem.
Book a 30-minute live read: describe one problem and see the likely causes ranked before the call ends.