01
Design your study
Tell the study-design agent what you want to find out. It turns a vague goal into a sound research question and audience, and flags weak designs before you spend a run.
Ask a question, pick who should answer it, and get a full report from AI twins built on real people — in minutes, without going into the field.
No panels to recruit, no fieldwork to schedule. Set up a study in the morning and have the report before lunch.
Tell the study-design agent what you want to find out. It turns a vague goal into a sound research question and audience, and flags weak designs before you spend a run.
Concept, price, message, comparison or open research. Add your product context and images so twins react to the real thing.
Choose a market and narrow by age, income, location and gender. Add screener questions and watch the audience estimate update live.
Review, add follow-up questions if you like, and run. Sentiment, objections, verbatims and segment cuts land when the run finishes.
Every study runs the same way, whatever the question. Ask in plain language and read a statistically grounded answer.
"At ₹1,499 a month, would you subscribe?"
"Which of these three pack designs would you pick up?"
"Which tagline makes you trust this brand more?"
"Does the festive offer or the everyday-value ad land better?"
"Which markets are most ready for this category today?"
"What is the one reason you would not buy this?"
AI twins answer instantly. What used to take weeks of recruiting and fieldwork becomes a same-day decision.
Define exactly who answers — niche audiences, specific cities, particular buying habits — without hunting for a panel that fits.
Change one assumption and run it again. Compare price points, messages or audiences side by side.
Each AI twin carries a full profile, so an answer comes from a specific individual with their own habits and constraints. Ask across six markets and 15+ industries.
Benchmarks are run against real-world outcomes, and we publish the results, including the parts that fall short.
Every twin answers on its own. A segment is only a filter, never one averaged persona speaking for a group.
Profiles come from real people, with personality, attitudes and buying behaviour, so answers reflect how people actually decide.
We test results against real-world data and publish where they hold and where they do not.
Open a simulation to see each twin's reaction, the objections that repeat, the quotes worth sharing and how segments differ. Chat with any twin to dig deeper.
"For two people that works out cheaper than ordering in twice a week. I would try it for a month."
"Depends on portion sizes and whether I can pause a week when I travel."
"My family cooks fresh every day. I would not pay this when I can buy vegetables locally."
A general chatbot gives you one averaged answer. Codon runs many independent AI twins, each built from a real profile, so you get a spread of opinions across segments rather than a single guess.
No seat fees and no long contracts. Start free and scale as your research grows.
1 simulation, 10 AI twins
400 AI twins a month
1,200 AI twins a month
Bulk volume and support
Codon is in private beta — AI Twins grounded in real survey data across India, the US, UK, Germany, France, and Italy. Come test it on your next decision.