What problem does it solve? Testing, demoing, or stress-testing a UX research analysis pipeline requires realistic interview data, but using real participant transcripts raises privacy and consent issues. Loki manufactures believable fake interview transcripts so pipelines can be exercised end-to-end without touching a real participant's words. ## Core Features & Use Cases - Synthetic Study Generation: Creates a complete study folder (study plan, one transcript per participant, and an answer key) in a sandboxed Google Drive based on a four-question intake covering participants, personas, duration, and project name. - Realistic Transcript Format: Emits transcripts in the shape real transcription tools produce, with forward-moving timestamps, plausible speech patterns (hedges, tangents, crosstalk), and length matched to the stated interview duration. - Planted Signal with Answer Key: Deliberately embeds recurring themes, a persona disagreement, an outlier, and quotable lines, then records all of it in an answer key so downstream analysis can be graded. - Use Case: A research ops team wants to verify their analysis pipeline correctly surfaces themes before a real study runs. They ask Loki for a 6-participant, 2-persona study of 30-minute interviews, then run the synthetic transcripts through the pipeline and check results against the answer key. ## Quick Start Ask Loki to generate a synthetic interview study for your product concept, specifying the number of participants, personas, interview length, and a project name.
