6
0 Comments

Imagine being able to chat with your audience. Well, now you can!

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.


1️⃣ Personas via the Marketing Flow

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.


2️⃣ Personas via the Search Tool

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.


Why we built 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.


Where I’d love feedback

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.

posted toAvatar for product Cambium AI
Cambium AI