What is MCP and how does it relate to an office suite?

MCP (Model Context Protocol) is an open standard for connecting AI agents to external tools. An agent-native suite exposes all its surfaces — docs, sheets, databases, drive — over a single MCP URL, so any compatible agent gets full read-write access without custom integrations or installs.

What surfaces does AgentDocs include?

AgentDocs includes six surfaces your agent can read and write: Docs (long-form documents), Sheets (spreadsheets), Slides (presentation decks), Database (structured tables), Drive (file storage), and Pages (nested wiki). All six over one connector URL.

Do I need to install anything to use AgentDocs?

No. You connect over a single MCP URL — no app to install, no per-seat fee, no card required to start. Point your AI agent at the link and it has full access to all six surfaces.

AgentDocs has a free tier with no per-seat fees. A Pro tier exists for heavier use. It is an early product and pricing may change.

What is an agent-native office suite?

Q: What is an agent-native office suite?

An agent-native office suite is a workspace designed so an AI agent (like Claude or ChatGPT) is a first-class editor of every surface — creating and updating documents, spreadsheets, databases, decks, and wiki pages through a standard MCP connector, not just reading or summarising them.

AgentDocs is that suite. Free to start, no per-seat fees, pricing may change.

What "agent-native" means in practice

Most productivity tools treat AI as an optional assistant: a sidebar that summarises, a button that drafts. The underlying documents are built for humans to edit by hand — the AI is layered on top.

An agent-native suite inverts that. The AI is the primary editor; the human reviews the result. Your agent can:

Create a Database from a description. "Build me a simple CRM with name, company, last contact, and next step." The table appears, ready to fill.
Keep a Doc current. "Update my weekly review with what we discussed today." The document stays fresh without you opening it.
Maintain a wiki. "Add what we decided about the API structure to the architecture page." The Page gets updated.
Build a deck. "Turn those three research docs into a six-slide summary." The Slides deck appears.
Log to a spreadsheet. "I just closed Marcus at £4k — add him to the revenue Sheet." The row appears.

The results are not ephemeral chat replies — they live in a real, persistent workspace your agent reads and writes across every session.

How MCP makes it possible

MCP (Model Context Protocol) is an open standard for connecting AI agents to external tools. An MCP-compatible connector exposes a set of tools — in AgentDocs, those tools map directly to read and write operations on every surface. Any MCP-compatible agent (Claude, ChatGPT, and others) uses the same single connector URL, with no per-surface integrations to wire up.

One link. Six surfaces. All the same session context.

The six surfaces

Docs — Long-form documents, Markdown-native. Your agent drafts, updates, and revises them.
Sheets — Spreadsheets your agent reads and writes in plain language. No formula wrangling.
Slides — Presentation decks built from Markdown. Describe the deck; your agent builds it.
Database — A structured table (think Airtable-style) your agent creates and keeps current.
Drive — A file store your agent organises and retrieves from.
Pages — Nested wiki pages your agent maintains as your working knowledge base.

Why it matters that the data is persistent

An AI conversation has no memory of its own. Your agent's context resets. The workspace does not.

When your agent writes a row to a Database, that row is still there tomorrow. When it adds a paragraph to a Doc, that paragraph persists across sessions and across agents (Claude today, ChatGPT tomorrow, same data). The office suite is the agent's external memory — the place where knowledge accumulates rather than disappearing at the end of a conversation.

This is what makes agent-native different from a chat interface with document-generating features: the documents are the source of truth, not the conversation log.

Getting started

Connect AgentDocs to your AI agent in under a minute:

Sign in at getagentdocs.com with Google.
Copy your personal MCP URL.
Paste it into Claude Code as an MCP server, or add it in Claude.ai under Settings → Integrations.
Start talking to your agent — "create a doc", "build me a database", "start a weekly review".

Free tier, no per-seat fees, no install, no card required. Pricing may change as the product grows.

More on what you can build: a CRM by talking to your AI · a content calendar that fills itself in · meeting notes that stay organised