An AI research engine that turns your documents, fieldwork and data into insights and recommendations, with a human in the loop.
Specialist agents carry the study through six stages. At four of them, nothing moves forward until a person weighs in.
{{ cur.ai }}
{{ cur.you }}
One link at a time, from what was said to what is recommended. Each quote is mechanically checked against its source text.
AI personas stand in for real participants, so a study can move through the pipeline without waiting on fieldwork. Simulated responses are always labelled, and real fieldwork replaces them when it arrives.
Test the instruments before going to the field. Get a critique of weak questions and a preview of the findings and recommendations they lead to.
Gradually let AI personas carry more of the fieldwork, so fewer studies wait on recruiting. Real participants remain the ground truth.
[One line: who ran the study and what they had to decide.]
[The result: what was accepted and what changed.]
Read the case study“[A verified quote from the study’s evidence chain.]”
Led to: [the recommendation it supports]Agents propose. A researcher approves at four checkpoints and makes the final call.
Recommendations trace back through insights, themes and codes to verbatim evidence, checked against its source text.
Responses from AI personas are marked at every step, and real fieldwork replaces them when it arrives.
Sample sizes come from established sample-size models, not the AI’s own guess.
External context comes from official announcements and data. Never rumours, leaks or speculation.
Projects stay on your device. AI runs locally by default, with cloud models opt-in.