---
title: One Hub for internal AI agents
description: Bring knowledge, skills and authorized tools into one Hub, then use that shared foundation through team chat, workflows and supported MCP assistants.
canonical_url: https://hubzoid.com/product
last_updated: 2026-09-26
---

# One Hub. Many ways to put it to work.

The product

Bring knowledge, skills and authorized tools together. Let your team use that foundation through chat, workflows and their existing assistants.

Try the Hub

Start with one Hub on your own laptop.

Add your team’s context, then share it once it proves useful. Or ask us to build it with you.

## Build context once. Keep making it useful.

A Hub is the shared foundation for a product, team or company. Keep instructions and knowledge understandable, maintain reusable skills, and connect the systems where authoritative data lives.

- Knowledge: terms, rules, decisions and reference material
- Skills: reusable instructions for doing a particular job
- Tools: authorized connections, scripts and MCP capabilities
- Ownership: someone responsible for keeping the foundation current

[Understand the Hub](https://hubzoid.com/docs/concepts/hub)

## A useful answer is only the beginning.

Teammates can ask about the business and request the next action in web chat, Slack, WhatsApp or Telegram. Bring relevant context into a familiar conversation, with access appropriate to the person and the channel.

- Use the context relevant to the question
- Bring sources into the conversation
- Keep action and review boundaries explicit

[Read the chat guide](https://hubzoid.com/docs/chat/web-chat)

## Give repeatable work a dependable home.

Define the task, connect the inputs and make the result easy to inspect. Use supported schedules and triggers for recurring work, and keep failures and review steps visible to its owner.

- Run a task on demand or through a supported trigger
- Inspect inputs, steps and outputs
- Decide what can happen automatically and what needs review

[Read the workflow guide](https://hubzoid.com/docs/concepts/workflows-and-schedules)

## Keep your tools. Add the shared brain.

Connect Claude Code, Codex and Hermes to the Hub’s context and tools through MCP. Each connection has its own authorization. Shared context does not mean shared credentials or identical conversations.

[Understand MCP connections](https://hubzoid.com/docs/guides/connect-an-assistant)

## Know what ran. See what needs you.

Give builders and owners one place to inspect agents, runs and team access. Follow a result back to its work, spot failures and make responsibility visible.

[Plan team operation](https://hubzoid.com/docs/console/overview)

## Control access down to the tool.

Decide who can use an agent and which tools they can call. Restricted tools are gated before execution. Agent access and tool permissions are separate decisions, and neither is made by the model.

[Read about access](https://hubzoid.com/docs/concepts/identity-and-access)
[Security responsibilities](https://hubzoid.com/security)