> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.moyoinnovations.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.moyoinnovations.com/_mcp/server.

# AI Assistant

> A private, streaming chat assistant for your team.

The AI Assistant module gives every team member a private chat assistant,
powered by Anthropic's Claude and OpenAI's GPT models, with persistent
conversation history, file attachments, and a choice of speed vs.
capability.

## Threads

Each conversation is a **thread**. Threads are private to the person who
started them — not shared with the rest of your organization, and not
visible to Owners or Admins either. This is different from every other
module in Moyo Suite, where an Owner or Admin can typically see everything
their team creates. A thread is treated the way a personal notes document
would be: people ask half-formed questions, paste drafts, and think out loud
in it, so nobody else in the workspace can read it, full stop.

A thread remembers your last 40 messages of context. Older messages still
exist in the transcript but drop out of what the assistant "remembers" for
new replies.

## Models

You choose a model tier per thread, not a specific model name — model
choice is deliberately abstracted so an underlying model can be upgraded
without changing anything you interact with:

| Tier             | Vendor | Best for                               |
| ---------------- | ------ | -------------------------------------- |
| **Fast**         | Claude | Quick questions, simple rewrites       |
| **Balanced**     | Claude | Everyday work — the default            |
| **Deep**         | Claude | Complex reasoning, multi-step problems |
| **GPT Fast**     | OpenAI | Quick questions, simple rewrites       |
| **GPT Balanced** | OpenAI | Everyday work                          |
| **GPT Deep**     | OpenAI | Complex reasoning, multi-step problems |

Every tier except **Fast** and **GPT Fast** supports **extended thinking** —
an optional toggle that lets the model reason step-by-step before answering,
shown to you as a collapsible "thought process" section above its reply.
It's slower and uses more tokens, so it's off by default. On the GPT tiers,
that section may sometimes stay empty even with thinking on — reasoning
still happens and is billed, but a written summary of it isn't always
available.

## Attachments

You can attach images, PDFs, and text files (plain text, CSV, Markdown,
JSON) to a message, up to 10MB each. Images are sent to the model directly;
text-based files have their contents extracted and included as context. PDF
text extraction is not yet available — a PDF can be attached, but its
content isn't read by the assistant yet.

> **Note**
>
> Attachments can only be added to a message inside an existing thread. The
> very first message of a new conversation is text-only.

## What leaves your workspace

Message text and attachment content are sent to Anthropic's API to generate
a response. Moyo does not train any model on your data. If your organization
has data-residency or processor-agreement requirements around third-party AI
providers, talk to your account team before enabling this module — it is
opt-in per organization, like every other module.

The assistant has no access to the rest of your Moyo Suite data (CRM, billing,
inventory, expenses) in this release — it can only see what's in the
conversation itself.

## Related

* [Plans & Modules](/guides/plans-modules) — how the AI Assistant is priced and included
* [Team & Permissions](/guides/team-permissions) — the permissions that gate sending messages and managing attachments