Contact by email
Back to Blog
AI News

A prompt set you will still understand next quarter

GlasBox Studio · · 3 min read

TL;DR: Key Takeaways

Monitoring without a fixed set is anecdote. A set has types, not a hundred variants.

  • - Type A: own brand plus service.
  • Each row: language, system, date, raw answer, class, visible source, note.
  • SOURCE/01 by GlasBox (AI) produces the first set from the intake call.

Monitoring without a fixed set is anecdote. A set has types, not a hundred variants.

- Type A: own brand plus service. - Type B: service plus place, no brand. - Type C: boundary ("no night work", "B2B only"). - Type D: comparison (two named competitors, fair and short). - Type E: fault test (a deliberately withdrawn service).

Each row: language, system, date, raw answer, class, visible source, note. Do not tune prompts "until it looks good". The set stays stable or you lose the time series.

SOURCE/01 by GlasBox (AI) produces the first set from the intake call. Change it only when the offer changes - and then version it.

---

An AI-visibility baseline

Next step

GlasBox measures classic SEO and AI visibility separately. No ranking or mention guarantee. The free short check stays thin on purpose.

Request an introductory call

Frequently Asked Questions

Why a fixed prompt set?
Monitoring without a fixed set is anecdote. The set stays stable or you lose the time series.
Which types belong in the set?
Own brand plus service, service plus place, boundary, short comparison, fault test with a withdrawn service.
When may the set change?
Only when the offer changes - and then version it. Do not tune prompts until they look good.