# AI Commerce

AI Commerce is Finqu's platform for building intelligent merchant experiences and automation in the admin. As a partner or developer, you can use these building blocks to create support tools, repeatable workflows, and background agents—all integrated with the same commerce core your APIs and storefronts use.

## Who it's for

- **Partners and agencies** building custom solutions or white-label offerings for merchants
- **Developers** extending the admin with chat, automation, or reporting
- **Product teams** shipping AI-powered features inside apps and integrations

## What you can build

| Capability | What it enables |
|------------|-----------------|
| [Success Agent](./success-agent) | Merchant-facing chat in the admin, support tools, and guided workflows. Use ask/agent/planning modes, subagents, and custom skills to build smarter admin UIs. |
| [Playbooks](./playbooks) | Step-by-step instructions for Workers (each step = a Worker doing a task). Run manually or as bulk actions—repeatable workflows you can ship to merchants. |
| [Guidelines](./guidelines) | Consistent agent behavior, brand voice, and guardrails. Customize how agents behave across stores or white-label solutions. |
| [Workers](./workers) | Background automation, reporting pipelines, and optimization loops. Use Executor, Planner, and Analyst as building blocks—schedule them, give them custom prompts and tools. |
| [MCP Tools](./mcp-tools) | Connect any MCP server to Finqu (OAuth, bearer, SSE, Streamable HTTP). One integration powers both Workers and the Success Agent. |

## How the pieces fit together

- The **Success Agent** is the conversational surface merchants use in the admin; it can call **Workers** and use **MCP Tools** you configure.
- **Workers** (Executor, Planner, Analyst) run on a schedule or on demand. **Playbooks** define steps that assign tasks to Workers; when the merchant runs a playbook, those steps are carried out. Workers also use **MCP Tools** for external data and APIs.
- **Guidelines** shape how all agents reason and respond, so you can align behavior with your or your merchants' brand and policies.

You can combine these to build end-to-end flows: for example, an Analyst report feeding a Planner that produces tasks, a playbook whose steps direct the Executor to apply changes, and a Success Agent that lets the merchant inspect or override—all using the same MCP tools and guidelines.

## Related: Agentic Commerce

For **customer-facing** AI that interacts with the store—shopping assistants, agent platforms, and multi-channel agents—see [Agentic Commerce](/agentic-commerce/overview). Agentic Commerce uses the [Universal Commerce Protocol (UCP)](/reference/ucp) to expose stores to external agents.

## Next steps

- [Success Agent](./success-agent) — Build merchant chat and support experiences
- [Playbooks](./playbooks) — Define step-by-step workflows for Workers
- [Guidelines](./guidelines) — Set rules and tone for agents
- [Workers](./workers) — Background agents: Executor, Planner, Analyst
- [MCP Tools](./mcp-tools) — Connect external MCP servers
- [Agentic Commerce](/agentic-commerce/overview) — Build AI-powered shopping experiences via UCP