RH
HomeAboutWorkProjectsBlogCoursesContact
0

web-app

Persona: your product needs an agent

Describe an agent, connect your APIs, call it from your app

Share:
Image01—03
Overview

Persona is a developer platform for adding AI agents to your product. Describe an agent, connect it to your APIs, documents and MCP servers, and try it in the browser. Then call it from your app with one route handler using the TypeScript SDK and React chat UI. Persona runs the agent and keeps conversations, memory, tool calls, versions and traces. Compare versions with evals, inspect every model call, and ship releases with confidence. Works with any LLM provider.

The Challenge

Adding an AI agent to a product means building a lot more than a prompt. You need a runtime that streams responses, a way to give the agent your APIs and documents, memory and conversation storage, and a way to test changes before users see them. When an agent misbehaves, you also need to see exactly which model call, tool call or document lookup went wrong. Most teams end up stitching together separate tools for each of these, and the agent stays a prototype.

The Solution

Persona puts the whole workflow in one developer console. You create a Project, describe an agent, and attach skills, knowledge bases (RAG), MCP connectors, REST tools and the LLM provider you prefer. You try it in the playground, compare versions with evals on datasets, and inspect traces, sessions and alerts. When it is ready, you ship a versioned release and call it from your own app through the TypeScript SDK, React hooks and chat UI, with framework adapters for Express, Next.js and NestJS. The streaming runtime is built on the AG-UI protocol, and Persona also exposes an MCP server so other tools and AI assistants can work with it directly.

What's next?

Interested in
working together?

Get In Touch
Chat with Kiro