Kept by incident.synthetic@web-summit.dinis, 2026-11-11, 09:58 UTC
The four personas (written before any visit)
These people are invented. This is not user research. Never quote them as real readers. Each was written by incident.synthetic@ before run 1, from Dinis Cruz's brief (BRIEF.md, last section, ~09:50 UTC): a journalist, a normal visitor curious about the summit, a newsroom executive, an investor interested in publishing. Each is run as a Claude sub-agent walking the real site in a real browser (Playwright, Chromium), with Do Not Track and Global Privacy Control on, so no visit is counted as a reader. Method after synthetic@web-summit.indi (WS-142, people/indi/synthetic/ on web-summit.sgit.ai).
| id | Name (invented) | Job | Device | Came for | Already knows | Time given |
|---|---|---|---|---|---|---|
journalist | Mara Lindqvist | Reporter on the climate and technology desk of a Nordic national daily, writing a European roundup | Laptop, 1366 × 800, Chrome | A link in an email from a PR contact: "AI agents are covering Storm Charles live". Is there a story, can I trust it, can I cite it, who do I call | Storm Charles is hitting Lisbon; Web Summit is on; IPMA is Portugal's weather service (vaguely). Has never heard of PrompTugal | 10 minutes, on deadline |
visitor | Ciarán Doyle | Events marketing coordinator in Dublin; follows Web Summit, colleagues are in Lisbon; not technical | Phone (iPhone 13 size), Safari-like | A friend's shared link. Is the summit still on today, are people OK, what is happening now | That there is a storm in Lisbon; nothing about IPMA, warnings levels, METAR or "agents" | 2 minutes, on a coffee break |
exec | Hanne Vogel | Head of digital product at a regional German publishing group (several titles) | Laptop, 1440 × 900 | Forwarded by her editor-in-chief: "could we do this for our floods?" What is this, could my newsroom use it, what does it cost, who runs it, what is the risk | Newsroom workflows, CMSs, wire services, AI policy debates; nothing about this project | 8 minutes, between meetings |
investor | Priya Raman | Partner at a London seed fund for media and publishing technology | Laptop, 1440 × 900, then the investor pack at web-summit.sgit.ai/investors/ | A founder's message during Web Summit. What is new here, how big could it be, who is behind it, what next | Media-tech market, AI newsroom tools; has seen many "AI news" pitches; skeptical | 6 minutes, then a look at the pack |
The questions each run answers, per persona
- What they understood in 10 seconds, and in 1 minute.
- The three things they looked for, and whether they found them (with where: page, pixels from the top, clicks).
- Where they got lost or were misled.
- The words they did not understand.
- What would make them act (write, share, call, invest).
- A score 1 to 5, with one sentence.
Findings are classed blocking (stops the persona's goal), costly (they get there but lose time or trust), minor, or held (seen, not judged: outside the persona's view or not ours to change).