SYNETHNO API

SYNETHNO API creates synthetic ethnographic agents from real research data, letting teams interview user segments to test hypotheses before fieldwork. Faster, cheaper, smarter UX research.

Comment

Inspiration

SYNETHNO API was inspired by a recurring gap in qualitative research and UX work: the need to make high-stakes decisions before real field research is possible. Ethnographic studies are powerful, but they are slow, expensive, and difficult to iterate. As a result, teams often rely on assumptions or oversimplified personas.

We were inspired by the question: What if researchers could safely test their hypotheses against realistic user perspectives before going to the field — without replacing real people?

What it does

SYNETHNO API creates synthetic ethnographic agents grounded in real qualitative research data.
By uploading interview transcripts, field notes, and market insights, users can generate a segment-level agent that reflects the language, concerns, and reasoning patterns of a real demographic group.

Researchers and designers can then “interview” these agents via API to test hypotheses, validate language, and uncover blind spots before conducting real-world research.

How we built it

The project was built as an API-first prototype focused on clarity, explainability, and ethical use.

  • Backend built with Python and FastAPI
  • In-memory data storage for rapid prototyping
  • Few-shot prompt grounding using real qualitative excerpts
  • Automatic API documentation via Swagger/OpenAPI
  • Developed and deployed using Replit

The entire prototype runs as a single file to enable fast iteration and easy demos during the hackathon.

Challenges we ran into

One major challenge was balancing realism with responsibility. The agent needed to sound human and grounded without pretending to be a real person. This required careful prompt constraints and explicit labeling of all outputs as synthetic.

Another challenge was scope control. Ethnography is deep and complex, but hackathons demand focus. We intentionally framed SYNETHNO as a hypothesis-testing layer, not a replacement for real research.

Accomplishments that we're proud of

  • Designing a novel application of AI for qualitative research
  • Creating a working, API-first prototype suitable for real UX workflows
  • Embedding ethical constraints directly into the system design
  • Translating a research-heavy concept into a clear, demo-ready product

What we learned

We learned that qualitative insights can be modeled not by mimicking individuals, but by capturing patterns of language, hesitation, and reasoning at the segment level.

We also learned that strong constraints improve AI behavior, and that API-first thinking helps turn abstract ideas into tangible products quickly.

What's next for SYNETHNO API

Next steps include:

  • Connecting the API to live LLMs for deeper grounding
  • Supporting larger and more diverse qualitative datasets
  • Adding evaluation tools to compare synthetic insights with real field results
  • Exploring integrations with UX research and design platforms

SYNETHNO API aims to become a trusted pre-field research layer that helps teams ask better questions — sooner.

Built With

  • base44
  • fas-api
  • llm
  • openai
  • pydantic
  • python
  • swagger
Share this project:

Updates