Deep research in design:
From data overload to
robust guidelines
This article tackles a modern UX problem:
the challenge of
building AI assistants
The piece shows how Ergosign rebooted a client project to build reliable design guidelines for dialogue-based AI assistants. The team paired desk research and official AI guidebooks from Google, Microsoft, and Amazon with the deep research tools in Perplexity, ChatGPT, and Gemini, then consolidated every validated source in NotebookLM. Structured meta-prompting sharpened the queries, but human judgment stayed at the centre. Findings were filtered, tagged, and checked with the client, then shaped into a clear, traceable guide of summaries, principles, and do's and don'ts. The takeaway: AI can accelerate research dramatically, as long as critical thinking and source verification stay in human hands.
As editor, writer, and project manager of the AI campaign, I led this piece from concept to publication.