Blog
Thoughts and notes
Thoughts, ideas and technical notes from our work with product teams and AI agents.
- Jochen Ade4 min read
User feedback doesn't scale until recruiting does
How always-on recruiting, combined with podojo's user tests and AI interviews, lets teams scale up user feedback.
Read article - Stefan Haas16 min read
LLM wiki vs. RAG: what the research says about a memory for discovery
What studies of LLM wikis, retrieval and agent memory tested, what we would build for customer research and what we have learned in practice.
Read article - Stefan Haas5 min read
What should a user research agent do?
What we're learning at Bolt about building an AI user research agent on an LLM wiki, and about making shared knowledge work for a team.
Read article - Stefan Haas8 min read
The future of customer research
On AI in customer research, continuous discovery and the work of keeping shared knowledge useful.
Read article - Stefan Haas17 min read
Everyone can become a knowledge builder
Two workshops, one pattern: non-technical teams built working LLM wikis with Claude Code in ninety minutes. The entry barrier has fallen. What matters now is designing the knowledge system the agent works in, not writing a better prompt.
Read article - Julian Siemer3 min read
From manual work to automated research recruitment
How Bolt's research team replaced a stack of disconnected recruiting tools with one composed workflow — and why the people closest to the problem built it themselves.
Read article - Stefan Haas10 min read
What my father the diagnostician knew about product taste
My father trained his diagnostic intuition by standing next to thousands of cases. Product taste works the same way — and that is exactly why it can be taught.
Read article - Stefan Haas4 min read
How to test prototypes with agentic AI
How agentic AI runs the whole prototype-testing loop: map the business model, shape a Test Card and a Lovable prototype, run an unmoderated user test, and synthesize the results.
Read article - Stefan Haas7 min read
Discovery's oldest bottleneck, meeting the newest way of building
On continuous discovery, why recruiting your own users can beat buying a panel, and a third way to build the internal tools you need.
Read article - Stefan Haas1 min read
AI Camp Berlin 2026 — Session Decks
Slide decks from our AI Camp Berlin session on agentic user research — the Podojo workflow and Ayse Delikaya's mobile.de results.
Read article - Stefan Haas2 min read
UX Testing with Claude Code — Meetup Recording
Recording from our April 21 Podojo meetup: an end-to-end UX-testing workflow inside Claude Desktop, and what 'agentic' actually means in practice.
Read article - Jochen Ade5 min read
Agentic User Research
A preview of our AI Camp Berlin talk — notes from our user research pilot with Bolt, what Claude Code did well, and where the agent still lags a human expert.
Read article - Stefan Haas6 min read
The new product development cycle: morning idea, evening feedback
How Gamma and Anthropic validate product ideas in hours, not weeks — and why your make vs. buy decisions just changed forever.
Read article