Lumenbase exposes a few ways in:

An agent can read a deal, update its stage, create a task, and log a note. All through the API, no clicking.

• REST API (`v1-public-*`): contacts, companies, deals, tasks, activities, invoices. Auth with API keys (`lmb_*`) and scopes.
• Agent API (Openclaw): auth with a personal token (`x-api-token`).
• Remote MCP: Bearer or OAuth, at `https://lumenbase.io/api/mcp`.
• Outbound webhooks: signed payloads when records change.

The Openclaw agent setup gives tools an AI can call directly:

So an agent can ask what it's allowed to do, then do it.

• `discover_tools` to see what's available
• `execute_tool` to run an action
• `today_briefing` to pull the daily list

Here's the loop in practice:

The pipeline stays in sync because every write goes through the same shared logic.

1. Agent reads the current pipeline through REST or MCP.
2. It spots a deal that needs a follow-up task.
3. It creates the task and updates the deal stage.
4. A webhook fires so your other systems hear about it.

Revenue operators and builders at $2M to $30M ARR companies. Teams that want to wire their own agents or scripts into the CRM. People who think in systems and want the data model to behave like one.