We’ve been working on a feature at Cambium AI that lets you chat directly with structured personas.
The key difference: these personas aren’t invented from scratch. They’re generated from structured public data (U.S. Census / ACS), then turned into usable audience simulations.
There are two ways to generate them.
This is built for founders and marketers.
You start with context:
What are you building?
Who do you think it’s for?
What problem are you solving?
Price point?
The system then:
Suggests likely population segments
Builds personas grounded in real demographic distributions
Lets you open a live chat with them
You can ask:
Would this pricing feel expensive?
What would make you trust this product?
What would stop you from buying?
Does this landing page headline make sense?
This is useful early — before you have user data — when you’re still shaping positioning and ICP.
The second path is more direct.
Instead of starting with an idea, you define the audience yourself — via chat.
For example:
“Working parents earning $70k–$120k in urban areas.”
“Recent graduates in their 20s renting, with student debt.”
“Small business owners with 1–10 employees.”
You describe the traits in natural language.
The system interprets that into structured demographic filters behind the scenes, builds a population-backed persona, and then lets you chat with them.
This is less about geography and more about audience definition.
It’s useful when you already have a clear ICP in mind and want to pressure-test it.
We kept seeing the same pattern:
Founders guessing their ICP
Messaging based on intuition
Paid ads launched without understanding segment size
Surveys with no sampling logic
We wanted something in between raw Census tables and pure AI roleplay.
Just structured public data turned into a usable simulation.
If you’re building:
Would you use personas before running ads?
Would you start with idea → persona (marketing flow) or population → persona (search)?
What would need to be true for you to trust this in your workflow?
Happy to share access in exchange for honest feedback. We’re still iterating.