How to build a Persona Feature Tester with Claude Code and ElevenLabs
🧠Bring your personas to life, run new ideas past them - and call them up to clarify any questions. Not a replacement for speaking to humans - but a quick way to challenge your thinking.
🔒 The Knowledge Series breaks down emerging AI technologies with practical playbooks designed specifically for product teams. Get 100+ guides and practical tutorials covering everything from Claude Code and MCP to agentic workflows, vibe coding, and more.
Many of the world’s leading product teams are shipping new features faster than ever. So far in April alone, Google has shipped 55 new releases, Notion has shipped 11, and Anthropic has shipped 21.1
Anthropic’s lead engineer, Felix Rieseberg, recently said that the speed of development means their roadmap is now just one month long.
But just because you can ship something doesn’t mean that your users want you to. Personas are a powerful way for product teams to remind themselves of their core user groups.
In this Knowledge Series, we’re going to take that principle and build something that will help you remember who you’re building a product for and why. The Persona Feature Tester will bring your user personas to life, let you describe a feature you’re building, or import it directly from Linear, and run it past your user personas all at once.
This is not meant to replace speaking to real users. Instead, it’s meant to be used as a tool to bring your user personas to life, test your assumptions and think about features from multiple different perspectives in one go.
Here’s what we’ll build together:
The persona feature tester is built using real world data sets and it allows you to describe a new feature or roadmap item and run it past multiple personas - all at once. If you have any further questions, you can call up a user persona and have a conversation:
For our example, we’ll imagine we’re working at a HR SaaS company called Kova but you can tweak this according to your own needs.
The app comes with a bunch of core features built into it including:
5 personas built using a mix of different types of real world data
The ability to describe your feature in text or import issues directly from Linear
A summary of feedback from each user persona, broken down by what they like and what they don’t
The ability to call up a user persona and have a conversation with them about the feature you’re planning to build
Personas are pretty divisive with some people arguing that they’re sometimes a novelty that’s not truly reflective of real world users. There is some truth to that, and they’re not a replacement for speaking to real humans, but for this, we’ll make sure each of our personas is fully grounded in real world data sets to try to make them as realistic as possible.
Getting set up
For this tool, we’re going to use a combination of some of the latest products to bring it to life, including: Claude Design, Code and Cowork, Cursor and ElevenLabs.
You’ll also get a downloadable version of the app and all of the examples of the real world data and folder structures that you can copy directly and hook up to your own instance of Claude Code.
There are 3 core parts to building this:
Creating our personas grounded in real world data with Claude Cowork and Projects
Building the first iteration of the app in Claude Design and Code
Hooking it to third party APIs including Linear, ElevenLabs and Anthropic to bring it all to life