# Overview

Missive has a built-in AI assistant that works directly inside your inbox. It reads your conversations, searches across all your connected accounts, checks your calendar, looks up contacts, and drafts replies. All without switching apps.

Setup is flexible. Start in one click with **Missive AI credits**, or **bring your own API key** for provider-side control. Either path works with **Anthropic Claude**, **OpenAI**, and **Google Gemini** -- pick the provider and model you want per task.

{% hint style="info" %}
AI requires the **Productive** or **Business** plan. Organizations on those plans start with **$5 of free Missive AI credits**. [See pricing](https://missiveapp.com/pricing).
{% endhint %}

## What the assistant can do

The assistant is not a simple chatbot. It has access to your Missive data and can take action on your behalf:

<table><thead><tr><th width="240.032958984375">Capability</th><th>How it works</th></tr></thead><tbody><tr><td><strong>Search your emails</strong></td><td>Searches across all connected accounts (Gmail, Outlook, IMAP, etc.) with filters for date, sender, recipient, and attachments.</td></tr><tr><td><strong>Check your calendar</strong></td><td>Reads events from Google Calendar and Microsoft Outlook. Can check availability, find conflicts, and suggest meeting times.</td></tr><tr><td><strong>Look up contacts</strong></td><td>Searches your contact books by name, email, or phone number. Returns full contact details including organization, job title, and custom fields.</td></tr><tr><td><strong>Find canned responses</strong></td><td>Performs semantic search across your canned responses to find relevant templates, policies, or pre-written replies.</td></tr><tr><td><strong>Create and edit drafts</strong></td><td>Writes reply drafts directly in the conversation. Handles recipients, subject lines, and signatures automatically. Can update drafts after feedback.</td></tr><tr><td><strong>Ask clarifying questions</strong></td><td>When your request is ambiguous, presents multiple-choice questions to clarify before proceeding.</td></tr></tbody></table>

Beyond the assistant, Missive offers additional AI features:

<table><thead><tr><th width="239.9244384765625">Feature</th><th>Description</th></tr></thead><tbody><tr><td><a href="using-ai/mcp-integrations"><strong>MCP integrations</strong></a></td><td>Connect the assistant to external tools like Notion, Linear, Attio, Stripe, or your own MCP server.</td></tr><tr><td><a href="using-ai/prompts"><strong>Prompts</strong></a></td><td>Quick, reusable AI actions you can run on messages or drafts: summarize, translate, fix grammar, generate a reply.</td></tr><tr><td><a href="../advanced-features/rules/ai-rules"><strong>AI Rules</strong></a></td><td>Trigger AI on incoming messages automatically. Classify, label, route, or draft replies based on content.</td></tr><tr><td><a href="using-ai/instructions"><strong>Instructions</strong></a></td><td>Define organization-wide behavior guidelines so the assistant responds consistently for your whole team.</td></tr></tbody></table>

## Two ways to connect

You can connect an AI provider two ways, and you can use either or both at the same time.

**Missive AI credits.** Missive manages the provider connection for you. You pay Missive for usage from a single credit balance that works across every supported provider. No API key required, and you get per-user cost analytics inside Missive.

**Bring your own key (BYOK).** You create an API key with the provider (OpenAI, Anthropic, or Google) and paste it into Missive. Usage bills directly to your provider account at the provider's price.

Only **organization admins** can add or manage AI providers, and every provider is linked to the organization (not a single user).

|                            | Missive AI credits                                                        | Bring your own key                                         |
| -------------------------- | ------------------------------------------------------------------------- | ---------------------------------------------------------- |
| Setup                      | Click **Add provider** in **Settings** > **AI**                           | Create an API key with the provider, paste it into Missive |
| Who you pay                | Missive                                                                   | The provider directly                                      |
| Pricing                    | Provider price + **25% markup**                                           | Provider price, no markup                                  |
| Free to start              | $5 one-time for Productive and Business plans                             | Depends on the provider                                    |
| Usage visible in Missive   | Actions **and cost** per user                                             | Actions per user                                           |
| Cost monitoring            | **Settings** > **AI** in Missive                                          | The provider's own dashboard                               |
| When the balance runs out  | Admins get a low-balance email; AI stops at $0 until an admin tops up     | You manage billing and limits on the provider side         |
| Advanced provider features | Limited (e.g., requests are routed through Missive's US provider account) | Full access (e.g., OpenAI EU data residency)               |

If ease of use matters most, pick Missive AI credits. If the 25% markup is a concern, or you need advanced provider features like OpenAI's EU data residency, [bring your own key](https://missiveapp.com/docs/ai/setup).

{% hint style="warning" %}
Each provider has its own data retention and privacy policies. By default, most providers do not use API data for model training, but you should review your account settings. See the data privacy section on each provider's setup page: [OpenAI](https://missiveapp.com/docs/setup/openai#data-privacy), [Anthropic](https://missiveapp.com/docs/setup/anthropic#data-privacy), [Google Gemini](https://missiveapp.com/docs/setup/gemini#data-privacy). This applies to both Missive AI credits and BYOK.
{% endhint %}

If you don't connect any AI provider, AI features simply don't appear in Missive. There's nothing to disable.

## Supported providers and models

You can connect one or more providers simultaneously. The assistant lets you pick which model to use per conversation.

| Provider             | Assistant models                                        | Rule models            |
| -------------------- | ------------------------------------------------------- | ---------------------- |
| **OpenAI**           | GPT-5.2, GPT-5 Mini, GPT-5 Nano                         | GPT-5 Mini, GPT-5 Nano |
| **Anthropic Claude** | Claude Opus 4.7, Claude Sonnet 4.6, Claude Haiku 4.5    | Claude Haiku 4.5       |
| **Google Gemini**    | Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash Lite | Gemini 2.5 Flash       |

Anthropic Claude is what the Missive team uses daily. Claude excels at nuanced writing and careful reasoning, which makes it particularly strong for drafting customer-facing replies.

{% hint style="info" %}
Model availability may change as providers release new versions. Missive updates supported models regularly.
{% endhint %}

## Next steps

* [Set up an AI provider](https://missiveapp.com/docs/ai/setup)
* [Connect MCP integrations](https://missiveapp.com/docs/ai/using-ai/mcp-integrations)
* [Use the assistant](https://missiveapp.com/docs/ai/using-ai/assistant)
* [Create prompts](https://missiveapp.com/docs/ai/using-ai/prompts)
* [Define instructions](https://missiveapp.com/docs/ai/using-ai/instructions)
* [Automate with AI Rules](https://missiveapp.com/docs/advanced-features/rules/ai-rules)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://missiveapp.com/docs/ai/overview.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.