What Expected Parrot does
Expected Parrot builds tools for AI-powered research that let teams simulate customers and survey respondents using AI agents and large language models. Its open-source Python package, EDSL (Expected Parrot Domain-Specific Language), lets users design, conduct, and analyze surveys and experiments run against AI agents instead of, or alongside, human respondents.
Key capabilities
With EDSL, users can turn customer data into AI agent personas and run them through surveys, simulating hundreds of demographically varied respondents, testing question and pricing variations, comparing responses across different language models, and labeling large text datasets. Results can be compared directly against real human survey data to support reproducible research. The company also offers Polly, a no-code interactive app, alongside the code-based library.
Who it's for
Expected Parrot serves product managers, designers, researchers, and data scientists doing market research and social science work. Typical use cases include testing reactions to new features, evaluating pricing, and gathering early signal before committing to costlier conventional research. The web app suits non-technical users while the Python library gives data teams full control. The company is backed by Y Combinator.