◐AI Lab3 min read

Persona-guided synthetic journey evaluation.

An AI-assisted pre-check: the agent uses validated persona context to walk a journey and surface likely risks before moderated or unmoderated research with real participants. It complements human research — it does not replace it, and it does not produce usability evidence.

Stage: Internal pilot
Type: AI-assisted pre-check (synthetic)
Connects to: Persona Repository
Measures: 4 agent-behaviour signals

overview

Overview

Usability testing usually stalls on setup, not analysis: picking the right persona for a journey, writing a task script, recruiting or simulating a session, and only then finding out whether the flow actually works. By the time results come back, the design has often already moved on.

This tool closes that gap by making a persona-matched pre-check something that happens automatically, the moment a journey is ready to test — not something scheduled weeks later.

The right persona, the right tasks, tested the moment the journey is built — not weeks later.

how it works

How it works

01

Connect to the persona repository

The agent reads from a maintained repository of validated personas — the same ones design and research already use — rather than a generic or invented test profile.

02

Match persona to journey

As a journey or flow is being built, the agent identifies which persona (or personas) it's actually designed for, based on the journey's context and intent.

03

Run a task set

A representative set of tasks is run against the journey in that persona's context — the same core tasks a moderated session would cover — as a rehearsal for one, not a substitute.

04

Analyse and report

Results are compiled into a structured readout the design team can act on immediately, without waiting on a separate research cycle.

impact

What it measures

Agent task completionShare of tasks the agent finished in the persona context

Simulated step countSteps taken versus the intended happy path

Detected friction eventsPoints where the agent stalled, backtracked or guessed

Journey completionWhether the end-to-end flow could be completed at all

Impact on the design lifecycle

Instead of usability testing being a scheduled, occasional event gated by recruiting participants, it becomes a continuous check available the moment a flow is ready — and because the persona is matched to the journey automatically, testing reflects the audience the journey was actually designed for, not whichever persona happened to be convenient to recruit.

Limits and honest caveats

This produces synthetic evidence. An agent is not a user, and its output is not a usability finding.
Every signal is named for what it is — agent task completion, not human task success — so results cannot be mistaken for research with participants.
Known weaknesses: model bias, persona fidelity, no emotional response, and no substitute for accessibility testing with disabled people.
Accuracy against real moderated sessions has not been benchmarked; that comparison is the next step before any accuracy claim.
Stage is internal pilot: used on real projects by a small group, still being validated.