About Navay
We build synthetic persona technology that helps teams understand how diverse audiences will respond — to content, research questions, concepts, or systems — before they commit resources.
The Problem
Most decisions start as debates
Research timelines don't match project deadlines
The default when there's no audience data
Teams ship products, policies, and campaigns based on intuition. Good ideas fail because nobody asked the right audience. Bad ideas ship because research was too slow or too expensive.
Chorus
Our product, Chorus, uses synthetic personas — AI individuals grounded in real demographic and cultural data — to run structured research. You define the question. Personas respond from their identity. An analyst synthesizes the findings.
Creative & message testing
Upload an ad, landing page, or message. Get structured feedback on resonance, clarity, and cultural fit across audience segments.
Qualitative research
Run surveys and focus groups with persona panels. Explore attitudes, perceptions, and cultural dynamics without recruiting.
Concept & scenario testing
Describe a product concept or policy. Have multi-turn conversations with personas who represent your target audience.
AI system audits
Stress-test your chatbot or AI assistant by simulating diverse user interactions. Surface failure modes before your users do.
Synthetic personas are strongest for directional research, hypothesis generation, and pre-launch testing. They supplement — not replace — traditional research with real participants. We publish our validation methodology so you can judge the fit for your use case.
Validation
We replicate published research and benchmark against third-party creative-testing reads to check whether synthetic personas recover the same findings real participants and paid tools did. We publish what we match, what we miss, and why.
Two moderated focus groups with the same 16 personas from our interview pass recovered all 11 food-choice themes — and surfaced 4 additional frames from group dynamics.
Read study →Two focus groups with 16 personas matched to Kumar's exact criteria recovered all 3 processed-foods themes and surfaced 5 additional frames on top.
Read study →Our 12-persona panel landed directional agreement with System1's paid read — same dip-then-rise emotional arc, same 0:32 trust pivot as the Star Rating driver.
Read study →Want to see what Chorus produces?
What Makes Us Different
Speed
Results in minutes, not weeks
Diversity
12,000+ personas, each built from real government statistics, labor surveys, and regional research — not generic AI profiles. 10K from global sampling (Western markets overrepresented); 2K built specifically for Indian demographics. Custom panels from your own customer data on demand.
Independent responses
Each persona responds alone — no conformity effects, no room dynamics skewing the data
Transparent method
We publish findings, share validation data, and show where results diverge — not just where they match
The Founder
Manish Patel
Connect on LinkedInFounder
14 years in AI and machine learning, 23 years in tech overall. Before Navay, Manish co-founded a movie recommendation startup built around the emotion a viewer wants to feel — he led the AI and backend work. The startup ran 18 months before he exited and started over.
For 12 years before that, Manish was at Mindtree, where he founded the applied machine learning research team from the ground up. Earlier years at several engineering shops rounded out a long arc of building systems that try to understand how people actually think — not how they're supposed to.
Navay is his attempt to turn that work into something other teams can use. The validation studies on the Evidence page — peer-reviewed replications and independent reads against third-party creative scoring — are the current working examples of the method.
Writing on AI, synthetic personas, and related topics lives on Substack ↗.
Interested?
Tell us what you're working on. We'll figure out if Chorus is a good fit.