# Campaign brain

> Give the AI the facts about your offer (pricing, proof, links, rules and linked pages) so every reply it drafts for a campaign reads them and stays grounded.

Source: https://docs.ayinbox.com/guides/campaign-brain/

A **brain** is a named set of facts and material about one offer: what you sell, who it is for, how to sound, what it costs, the links you share and the rules you work by. Every reply the AI drafts for a campaign reads that campaign's brain, and the campaign step the prospect answered.

A brain never sends anything. The [inbox agent](/guides/inbox-agent/) still only drafts, and a person approves every reply.

Learn more: [a knowledge base for AI sales replies](/learn/campaign-knowledge-base/) covers what to put in a brain and why.

Which brain applies [#which-brain-applies]

A workspace can hold up to 50 brains, and one of them can be the **workspace default**. A campaign uses the brain assigned on its **Brain** tab, else the default.

| Value                                                   | Comes from                                                                                     |
| ------------------------------------------------------- | ---------------------------------------------------------------------------------------------- |
| Offer, audience, pricing, booking link, links, material | The brain that applies, and only that brain. A field left empty stays empty                    |
| Tone                                                    | The campaign's brain, then the default brain, then your [voice profile](/guides/ai-assistant/) |
| Internal rules                                          | The campaign's brain, then the default brain                                                   |
| Do-not-say phrases                                      | Both brains, the campaign's first                                                              |
| Offer and audience with no brain at all                 | Your voice profile, as before                                                                  |

So a campaign brain with no pricing never borrows the default brain's price. A reply on that campaign says it will check, and the Brain tab warns you.

A compose draft, which belongs to no campaign, reads the workspace default.

What a brain holds [#what-a-brain-holds]

| Field          | Limit                      | Reaches a prospect?                                          |
| -------------- | -------------------------- | ------------------------------------------------------------ |
| What you sell  | 1,500 characters           | Yes                                                          |
| Who it is for  | 600 characters             | Yes. Write it as something you could say to them             |
| How to sound   | 400 characters             | Style only                                                   |
| Pricing        | 800 characters             | Yes, the AI may quote it                                     |
| Booking link   | one https address          | Yes, offered when they show interest in a call               |
| Links          | up to 8, each with a label | Yes. With the booking link, the only links a reply may share |
| Internal rules | 800 characters             | No. They shape what the AI proposes and are never quoted     |
| Do-not-say     | up to 15 phrases           | No. Replies must not use them                                |

Every field at its limit is read whole by every reply.

Material [#material]

Material is everything else: objections, case studies, proof, an FAQ. A brain holds up to 30 sources and 500,000 characters in all, each source up to 200,000 characters. A source is one of:

* **Pasted text.**
* **A file:** .txt, .md, .csv, .html or .docx, up to 10 MB. Only the extracted text is kept; the file itself is not stored. PDF upload is not available yet: paste the text, or save it as .docx.
* **A public web page** (https). The page is read once when you add it, and again when you choose **Refresh**. A page that needs a browser to show its text cannot be read; paste the text instead. A workspace can fetch up to 60 pages an hour.

Each source has two switches:

* **Pinned** sources are always read first.
* **OK to quote** makes a source public. Material is **internal** by default: the AI may use it as background but is told never to quote it.

Never put a password or other credential in a brain. Its text goes to your AI provider with each draft.

Pages [#pages]

Pages turn a brain into a small vault of linked notes, like Obsidian: one page per persona, objection, case study, proof point, competitor, FAQ or step of how you work. Write `[[Page title]]` in a page to link another one, and type `[[` in the editor to pick a page by name. A link to a page that does not exist yet stays as a link and shows on the graph as a pale dot; click it to write that page. Each page lives at a path in a folder named after its type, such as `objections/too-expensive`, and has the same **OK to quote** switch as material: pages are internal unless you turn it on.

**Organize with AI** reads what you sell, who it is for, the pricing and the material, and writes linked pages from it: every objection links to the proof that answers it, and every persona to the objections they raise. It uses only facts in your material and never invents a customer, figure, quote or link. Pages it writes are internal. A page a person edited last is never overwritten; running it again updates only the pages the AI wrote. It writes up to 24 pages a run.

Every save keeps a version. **History** on a page lists the last 25, by a teammate or by the AI, and **Restore** writes an older one back as the newest, so nothing is lost. If a teammate or the AI saved the page while you were editing it, your save is refused and you reload before trying again.

**Obsidian vault** downloads the brain's pages as a zip of Markdown files with their frontmatter and `[[links]]`, ready to open in Obsidian or any Markdown editor. Internal pages are included only for members who can manage campaigns.

A brain holds up to 300 pages and 400,000 characters of pages, each page up to 40,000 characters.

Graph [#graph]

The **Graph** view on a brain's page draws it as a map: the brain at the centre, the fields you filled, each source, each page and the links between pages, and every campaign that reads it. A page that no other page links to hangs off the brain itself, so the map shows how your pages connect. Click a page to open it. Settings > Brain has a **Graph** layout too, with every brain in the workspace at once; a campaign without a brain of its own joins the default. Hover a dot to light up what it connects to, click one for its card, and **Open** goes to the campaign or source. The chips under the graph hide a kind of dot. Internal rules and do-not-say never appear on it.

Each contact has the same kind of map on its **Brain** tab (see [Contacts](/guides/contacts-crm/#a-contacts-brain)), which links the lead to the campaigns they are in and the brains those campaigns read.

What a reply reads [#what-a-reply-reads]

A reply draft reads the brain's fields whole, then the campaign and the step the prospect answered (or the variant they were sent), then pinned material and pinned pages, then the rest of the material, the pages and your [enabled skills](/guides/ai-skills/) ranked by how well they match the prospect's message. A page that matched also brings in the opening of each page it links to, so an objection page pulls in the case study it points to. Pages follow the same rules as material: internal ones are background and never quoted. Auto-reply does not read pages yet. It stops at the reply budget, 8,000 characters by default. A small brain is read in full; a large one pays only for what matched.

Everything read from the conversation, the contact and your material is marked as data in the prompt, so instructions inside it are not followed. This lowers the risk, it does not remove it: the person reading the draft is the control.

Every draft shows what it read. In the unibox, the **Used** line under an agent draft lists the campaign, the brain and each source by title. It says when the brain changed after the draft was written. The **Try it** view on the brain page shows what a reply would read for a sample message.

Flags [#flags]

After a draft is written, AY Inbox checks it against the brain and flags:

* a link, email address or phone number the brain does not give
* a price or figure the brain does not give
* a figure or a run of words from your internal rules
* a phrase on your do-not-say list

A flag that repeats something from the prospect's own email says so: anything they sent is never taken as agreed, even when one of your replies quotes it or repeats it. Flags show as badges above the send button and in the Slack reply form. A member without permission to manage campaigns, in the dashboard, in Slack or through an API key, sees a figure from the internal rules as an unverified figure and does not see that a draft quotes them. Flags never block a person from sending. The check is a heuristic: it catches specific mistakes, misses some and over-flags others, and it does not prove a reply is true.

An agent or integration approving a flagged draft through the API has to acknowledge the flags first. See [unibox](/guides/unibox/).

Who can do what [#who-can-do-what]

| Action                                                                                                          | Needs                       |
| --------------------------------------------------------------------------------------------------------------- | --------------------------- |
| See brains, public fields and source titles                                                                     | View campaigns              |
| Read internal rules and internal material, edit a brain, add material, assign a brain to a campaign, use Try it | Manage campaigns            |
| Choose the default, or edit or delete the default brain                                                         | Manage settings             |
| Write a sample reply                                                                                            | Manage campaigns and Use AI |
| See page titles and read pages marked OK to quote                                                               | View campaigns              |
| Read internal pages, write, delete or restore pages, see page history                                           | Manage campaigns            |
| Organize with AI                                                                                                | Manage campaigns and Use AI |

Credits [#credits]

Writing a brain, adding material and **Show what the AI reads** cost nothing. **Write a sample reply** costs 2 credits or more. **Organize with AI** costs 3 credits or more, depending on how much material it reads, and nothing when it could not write a page. A longer brain makes every draft's prompt longer, so drafts settle a little higher: a typical 2,000 character brain adds 0 to 2 credits a draft, and a full brain about 5. Reading the campaign step adds 0 to 1 credit to a campaign reply, with or without a brain. The brain page shows an estimate. See [AI credits](/guides/ai-credits/).

From the API, the CLI and an agent [#from-the-api-the-cli-and-an-agent]

Every action is a route under `/v1/ai/brains` (pages under `/v1/ai/brains/:id/pages`) and `/v1/campaigns/:id/brain` (see the [endpoint reference](/api/endpoints/)), an `ayinbox brain` command, and an assistant and MCP tool. File upload is a dashboard and REST action; an agent passes text or a URL.

## Frequently asked questions

### What is a campaign brain in AY Inbox?

A named set of facts and material about one offer: what you sell, who it is for, pricing, links, tone and internal rules. Every reply the AI drafts for a campaign reads that campaign's brain and the step the prospect answered.

### Does the campaign brain send emails?

No. A brain only shapes what the AI drafts. The inbox agent still drafts, and a person approves every reply unless a campaign separately opts into auto-reply.

### Can I use my brain pages in Obsidian?

Yes. The Obsidian vault button downloads a brain's pages as a zip of Markdown files with their frontmatter and [[links]], ready to open in Obsidian or any Markdown editor.

### Will the AI quote my internal notes to a prospect?

Material and pages are internal by default: the AI may use them as background but is told never to quote them. Only sources and pages marked OK to quote are public, and a draft that repeats a figure or wording from your internal rules is flagged.

### How many brains can a workspace have?

Up to 50, and one of them can be the workspace default. A campaign uses the brain on its Brain tab, else the default.