Agentic Eyes personas

A persona is the simulated customer’s character sheet: who they are, what they want, how patient they are, how much they trust you. It drives every judgment the review makes — the same site reads very differently to a cautious planner on her phone than to a hostile customer hunting for flaws.

Built-in personas

Sally — the Dreaming Planner

A 38-year-old planning a once-in-a-lifetime trip. High intent but cautious with money. Mobile-first, skims rather than reads, hunts for prices and trust signals. Won’t create an account just to look around. The default persona — she converts when a site earns trust, and her report reads like growth feedback.

The Worst Customer Ever

The adversarial twin: impatient, zero benefit of the doubt, looking for reasons to leave. Where Sally reports friction, the Worst Customer reports dealbreakers. Run both — the gap between their experiences maps your audience’s tolerance band.

Custom personas

The persona library lets you save your own: name, goal, patience level, tech comfort, language, device, trust posture, reading style, and a list of quirks — the specific behavioral notes (“reads every footnote”, “abandons at forced signup”) that make a persona feel like a person rather than a slider set. Custom personas get a stable id you can reuse across runs and share with your team.

How persona attributes shape runs

• Goal + patience bound the journey: a low-patience persona abandons sooner and reports why; the same friction from a high-patience persona reads as a minor note.
• Trust + reading style tune what counts as a finding: cautious skimmers flag missing prices and policies; thorough readers flag contradictions.
• Device sets the browsing context for layout and touch findings.

Personas compose with every other capability: consensus panels run your persona across models, and drive mode turns any persona into a clicking visitor rather than a first-impression critic.