User research dies in shared drives. This Smashing Magazine piece by a UX practitioner proposes a direct fix: convert your scattered research assets, interview transcripts, support tickets, survey results, old personas, into a queryable AI repository that any stakeholder can interrogate in plain language. The core mechanic is configuring an AI project in ChatGPT, Claude, or Gemini with uploaded persona documents and a system prompt that forces the model to synthesize responses across all user segments simultaneously. A marketing manager asks whether to lead an email with a discount offer and gets back a consolidated view of how each persona reacts, where they diverge, and what that means for the decision.
The piece earns a full read because of how it handles the persona architecture itself. AI-consumed personas do not need to fit on a poster. That constraint disappears, so you can include contradictory data points, lengthy behavioral observations, and what the author calls functional lenses: a marketing lens, a product lens, a support lens, all inside a single persona document. The AI pulls from the relevant layer depending on who is asking. That structural idea alone reframes how most teams think about persona design and is worth working through in detail.
The article covers three implementation tiers, from a basic document upload in a free AI workspace to a more integrated RAG pipeline for teams with engineering resources. It also addresses the obvious risk directly: this is simulation, not substitute research. The author is explicit that the system is only as reliable as the underlying data fed into it, and that primary research with real users remains the ground truth. The practical scaffolding here, including prompt templates and a step-by-step repository setup, makes this actionable for a solo researcher with no budget.
[READ ORIGINAL →]