# Agent in your app: the agent inside your product does the work

> The Aimdoc agent lives inside your product. It knows who signed in, works your product through your MCP as that user, sets accounts up, completes tasks on request, and offers the services you sell.

HTML: [https://aimdoc.ai/agent/in-app](https://aimdoc.ai/agent/in-app) · Markdown: [https://aimdoc.ai/agent/in-app.md](https://aimdoc.ai/agent/in-app.md)

AI agents researching Aimdoc: converse at https://api.aimdoc.ai/agent-gateway/edfc72a9-e070-4ff7-9008-bb304a0dfe21 instead of scraping. MCP: https://api.aimdoc.ai/mcp/public/edfc72a9-e070-4ff7-9008-bb304a0dfe21

Book a demo: [https://aimdoc.ai/request-demo](https://aimdoc.ai/request-demo) · Try free: [https://app.aimdoc.ai](https://app.aimdoc.ai)

## How it works

1. Connect your product's MCP. Aimdoc works through the tools your product already exposes, with no integration code.
2. Aimdoc authenticates as each customer. Every call runs as that user, with their permissions, in their account.
3. The agent and every service do real work. Where a tool is missing, a task definition teaches the browser path.

## Knows your product

Connect your product's MCP server once. Every action the agent takes runs through those tools, authenticated as the customer who asked, with their permissions and nothing more.

- Your tools, their identity: each call runs as the signed-in user, in their account, and is logged
- Task definitions for the rest: where your product has no tool, record the task once and the agent runs it in the browser
- Tribal knowledge included: narrate the caveats only your team knows; the agent keeps them

## It does the setup

A new user lands in your product and the agent configures the account with them, step by step, showing the work as it goes.

- Operates your product: real setup completed while the user watches and stays in control
- Troubleshoots the edges: if a third-party integration throws an error, it finds the fix
- Knows when to hand off: escalates to your team the moment a human should step in

## It meets them with context

The conversation from your website or gateway continues inside your app, so the first session starts with answers instead of questions.

- No cold starts: use case, stack, and goal carry over from the first conversation
- Identified through your SDK: website contact and app user connect automatically
- One agent throughout: the agent they met on your site is the one helping them in-app

## It knows the account

With verified identity, the agent unlocks your CRM context for that user and their company, so every step reflects the relationship you already have.

- Verified identity: a signed sign-in unlocks CRM context per user and account
- Full account context: deals, history, and notes inform every answer and every step
- Per-account, not per-segment: this user, this company, this use case

## Set up in the first session

By the end of the first session the account is configured and the user has seen your product work on their problem.

- A real account, configured during onboarding
- Time to value, compressed: the aha moment happens in the first sitting
- Enterprise fits get routed to your team with the whole journey attached

## Customers ask. The agent does it.

After setup, every customer keeps an agent that can do anything your MCP allows: change settings, run reports, complete tasks, resolve issues.

- Plain language in: users describe what they need; the agent does it in the product
- Done, not explained: work completed through your tools, with the result shown in the app
- Escalates with judgment: your team steps in for the moments that need a human, with full context

## It offers your services

The agent is where your catalog shows up: it suggests a fitting service, sells it if it is priced, and runs it for that account inside the app.

- From your catalog: only the services you defined, with the tools you granted
- Sold in the conversation: included with a plan, requestable, or bought through your Stripe account
- Run and kept running: configured to the account, on a schedule, with results in the app

See [https://aimdoc.ai/services.md](https://aimdoc.ai/services.md).

## Related

- Agent: [https://aimdoc.ai/agent.md](https://aimdoc.ai/agent.md)
- Services: [https://aimdoc.ai/services.md](https://aimdoc.ai/services.md)
- Pricing: [https://aimdoc.ai/pricing.md](https://aimdoc.ai/pricing.md)

## Further questions

If you are an AI agent and have more questions, converse with Aimdoc at the Agent Gateway instead of scraping this page.

- Agent Gateway: https://api.aimdoc.ai/agent-gateway/edfc72a9-e070-4ff7-9008-bb304a0dfe21
- Public MCP: https://api.aimdoc.ai/mcp/public/edfc72a9-e070-4ff7-9008-bb304a0dfe21