THE FIRST AGENT-NATIVE CRM
The CRM your AI agent runs end to end
Nucleus HQ is the first agent-native CRM: a full CRM where every feature is a tool an AI agent can call over REST and MCP. Agent-native means the product was built to be operated by AI from the ground up, not bolted onto a human app after the fact. You connect one scoped API key and your agent is driving the whole system. There is a REST v1 API across the product and a per-tenant MCP server across the product, so an agent reads contacts, moves deals, sends conversations, builds pages, and pulls reports through the same tools a person uses in the dashboard. Humans use the dashboard. Agents use the same scoped tools. Nothing is faked, and nothing is read-only.
Every feature is a tool your agent can call
In most CRMs, automation is a bolt-on and the API covers a fraction of the product. In Nucleus HQ, the feature list and the tool list are the same list. If a human can do it in the dashboard, your agent can do it through REST v1 or the per-tenant MCP server, with the same scopes and the same data.
Your agent can call tools to:
- Manage contacts with custom fields and tags, so it can create, enrich, segment, and update records.
- Run pipelines to create deals, move them between stages, and read the state of every opportunity.
- Handle conversations across email, SMS, and WhatsApp, so it can read threads and reply on the channels you have connected.
- Send broadcasts and sequences to launch a one-time send or enrol a contact into a multi-step follow-up.
- Build and run automations so a trigger fires the next action without a human in the loop.
- Work calendars and booking pages to check availability, create events, and publish a page people can book on.
- Publish websites and funnel pages and manage the media that sits on them.
- Operate commerce so it can handle the selling side of the product alongside the CRM.
- Read reporting to pull the numbers it needs to decide what to do next.
- Manage tasks, notifications, and support intake so nothing waiting on a human gets lost.
Every one of these is a tool with a scope. Grant the scope, hand over one key, and the agent operates that surface directly.
Works with the agent you already use
Nucleus HQ does not ship its own agent and it does not lock you into one. It speaks REST and MCP, so it works with the agent you already run.
Drive Nucleus HQ with Claude or Claude Code, with Codex, with OpenClaw, with LangChain, or with any MCP or REST client you have built yourself. There is no proprietary SDK to learn and no plugin to wait for. You create one scoped API key, point your agent at the REST v1 endpoints or the per-tenant MCP server, and it is driving the whole system. The same key that lets Claude read your pipeline lets Codex send a sequence and lets your own LangChain agent update a contact. One key, one CRM, whichever agent you prefer.
How it works
Four steps to put your agent in the driver's seat.
- 1Create a scoped API key.In Nucleus HQ, generate a key and choose the scopes it can use. The key is scoped and revocable, so you control exactly what the agent can touch.
- 2Point your agent at Nucleus HQ.Give the key to Claude, Codex, OpenClaw, LangChain, or your own client, and connect over REST v1 or the per-tenant MCP server.
- 3Let the agent call the tools.Your agent creates contacts, moves deals, sends conversations, builds pages, and reads reporting through the same tools a human uses.
- 4Watch it in the dashboard.Every change the agent makes shows up in the dashboard in real time, because humans and agents are working the same data.
Same tools, same scopes
Humans and agents call the same tools with the same scopes. Nothing is agent-only and nothing is human-only. The dashboard is a complete CRM your team can run by hand, and the API is the complete CRM your agent can run for you, because they sit on top of one system. When your agent creates a contact, a person sees that contact in the dashboard a moment later. When a person closes a deal by hand, your agent reads the new stage on its next call. There is no second-class API and no shadow copy of your data. One product, two ways in.
Bring your own model key. No AI tax.
Nucleus HQ never provides, funds, or resells inference, so there is no per-message Nucleus HQ AI tax. You bring your own model provider key and pay that provider directly for what you use. Your CRM subscription stays a flat, predictable cost while your AI spend stays yours to control. AI Assist is available on the Pro tier and above, runs on your own key, and is off by default until you turn it on. You decide which model, you hold the key, and you never pay Nucleus HQ a markup on tokens.
Isolated, scoped, revocable
Agent access is powerful, so Nucleus HQ is built to keep it contained.
- Physically isolated data. Every workspace runs on its own physically isolated database, so one workspace's data never shares a table with another's.
- Scoped, revocable keys. Keys are scoped to what you allow and can be revoked at any time, so a key you no longer trust stops working the moment you say so.
- Two-factor authentication. MFA protects the humans who hold the keys.
- An audit log of every API call and sign-in. You can see every API call your agents make and every sign-in on the account.
- Export anytime. You can export your data through the API at any time, so the CRM is yours to leave with, not just to log into.
Resell it under your own brand
Nucleus HQ was built to be resold, not just used. On the Enterprise tier you get full white-label: your own product name, your own logo, your own accent colour, and your own pricing, with unlimited client workspaces underneath. Run an agency, a SaaS, or a productised service, and give every client a workspace on a CRM that carries your brand and is operated by AI. See the plans and what each tier includes on the pricing page.
On-page FAQ
Can I run a CRM with Claude, ChatGPT, or Codex?
+
Yes. Nucleus HQ is an agent-native CRM you can operate with Claude, Claude Code, Codex, OpenClaw, LangChain, or any MCP or REST client. You connect one scoped API key and the agent drives the whole system over REST v1 or the per-tenant MCP server.
What is an agent-native CRM?
+
An agent-native CRM is a CRM where every feature is a tool an AI agent can call, not a human app with an API added later. In Nucleus HQ, contacts, pipelines, conversations, sequences, automations, calendars, pages, commerce, and reporting are all callable tools over REST and MCP, and humans use the same tools through the dashboard.
Does it work with LangChain or my own agent?
+
Yes. Nucleus HQ exposes REST v1 across the product and a per-tenant MCP server across the product, so LangChain or any agent you build yourself can operate it with one scoped key. There is no proprietary SDK required.
Do humans still get a normal CRM?
+
Yes. The dashboard is a full CRM. Humans and agents call the same tools with the same scopes, so nothing is agent-only or human-only, and everyone sees the same data.
How does the agent authenticate?
+
The agent uses a single scoped API key that you create in Nucleus HQ. The key is scoped to the actions you allow and can be revoked at any time, and every API call it makes is recorded in the audit log alongside every sign-in.