---
title: "How to give multiple AI assistants the same brand guidelines | Rainbrand"
description: "Connect assistants to the same maintained brand, give each task the guidance and files it needs, and track the snapshots that still need replacing."
url: "https://www.rainbrand.com/blog/give-multiple-ai-assistants-the-same-brand-guidelines/"
---

By [Tomas Ward](https://www.linkedin.com/in/tomasward1/)

Published 28 Oct 2026

6 min read

# How to give multiple AI assistants the same brand guidelines

Connect assistants to the same maintained brand, give each task the guidance and files it needs, and track the snapshots that still need replacing.

## TL;DR

Have each assistant retrieve guidance and files from the same maintained brand when brand work starts. Keep a short instruction explaining when to fetch it, and remove conflicting saved copies. Tools that need an export should record its version and refresh it after changes. Check what each assistant receives before relying on the setup.

## Share the brand, then tailor the task

Several assistants can connect to the same brand service and still use different product names. One may retrieve current guidance while another relies on a PDF uploaded before a rename. Sharing access is the first step; each task also needs to use that access.

Start with a shared reference for the brand. Then have each assistant retrieve the guidance and assets for its own task. A shared reference keeps a product rename or palette change from becoming a separate editing job in every assistant.

Write down which brand to use, when to retrieve it and how to handle saved notes that conflict. Those instructions make the shared reference part of each task.

This guide covers that setup. If you are already distributing a change, use the [update checklist](https://www.rainbrand.com/blog/update-brand-guidelines-across-ai-tools/).

## Establish which brand every assistant should use

Identify the maintained guidance and assets that the assistants should retrieve. Your team may already manage source documents in a brand platform or shared drive. Those sources can stay where they are; the important question is how a change reaches the context each task receives.

For recurring work, Rainbrand maintains shared brand context that agents can retrieve and update. It can work alongside existing sources. Connect each assistant through the [MCP quickstart](https://docs.rainbrand.com/mcp/getting-started). MCP, the [Model Context Protocol](https://modelcontextprotocol.io/docs/getting-started/intro), is a standard for connecting AI applications to external tools and data. Availability depends on the assistant and account, so follow the current setup for each one.

In each connected assistant:

1. Ask it to show the connected Rainbrand account.
2. List the brands it can access and select the intended ready brand.
3. Record that Brand ID, the identifier for the brand, in your setup notes.

For team work, check the team as well as the brand name. Rainbrand [adds a brand](https://docs.rainbrand.com/guides/create-brand) to a personal account by default; name the intended team when adding a shared brand. Connecting three assistants to three separately created brands will leave you with three brands to maintain.

You can keep a simple inventory:

| Workflow | Brand reference | How it gets context |
| --- | --- | --- |
| Email drafting | Selected team and Brand ID | Requests guidance for the email |
| Landing-page work | The same Brand ID | Requests layout guidance and files |
| A tool that only reads local files | Exported brand version | Reads a folder that needs refreshing |

## Give each assistant a rule for retrieving guidance

The reusable instruction should describe the retrieval step. Product names, colours and the rest of the brand guidance come back with the task context.

With Rainbrand, that task context is a [Brief](https://docs.rainbrand.com/guides/request-brief): selected brand guidance in Markdown, together with assets such as logos, fonts and reference images.

The following instruction is a template. Acme and Atlas Pro, its design software for product designers, are fictional. Replace Acme with your selected brand, and keep the Brand ID available to the assistant or your integration.

Add to each assistant’s instructions

```
For Acme brand work, use the selected Rainbrand brand. Before drafting or making an output, request a Brief describing the task, audience, channel and any exact copy. Wait for it to complete, then use its guidance and relevant files. Follow the current Brief over older saved brand notes and tell me about any conflict. Pass the guidance and files to any tool making the output. If the brand is unclear or retrieval fails, report that before continuing.
```

Put this in the instruction mechanism that the assistant uses for the workflow. Check that it is active in the actual conversation, project or repository where work happens. An instruction saved somewhere else will not help that task.

Remove the old brand summary once the connection works. Keeping both invites the assistant to combine current guidance with an obsolete product name or logo.

## Ask for the context each output needs

Sharing a brand does not mean every assistant should receive an identical Brief. Ask for the work you are about to make.

For a fictional Acme launch, the email request might be:

Ask your agent

```
Request a Brief for Acme’s Atlas Pro launch email, aimed at product designers. Use the headline ‘Meet Atlas Pro’ and the call to action ‘Explore Atlas Pro’. Show me the guidance before drafting the email.
```

The landing-page request would describe the page, audience and required sections. An image request should include dimensions, required copy and how the image will be used. The [Brief guide](https://docs.rainbrand.com/guides/request-brief) shows how to describe those requirements.

Inspect the returned context for the things the outputs should share: the product name, permitted claims, voice and relevant identity rules. Their selection and emphasis can differ because the tasks differ.

Then pass the original task, guidance and selected assets to the [tools creating the work](https://docs.rainbrand.com/guides/use-your-own-generation-tools). An agent that retrieves a logo and then asks an image model to redraw it from a description has lost a useful part of the context. A page renderer may need the actual font file; a model judging lettering may also need its specimen image.

Rainbrand supplies that context. The assistant, generation model or renderer still makes the output.

## Keep files together when a tool needs a snapshot

Some workflows need local files instead of a task-time connection. A [Brand Skill](https://docs.rainbrand.com/guides/use-brand-skill) is a complete, versioned snapshot of the brand's identity, guidance and asset references. Its exported folder includes the assets.

You can inspect the [TANNIN example package](https://www.rainbrand.com/downloads/tannin-brand-skill.zip). TANNIN is a tea-room concept; this package contains its instructions and assets. Start at `SKILL.md`, then follow `references/voice-and-messaging.md` and the relative asset links. The folder keeps the guidance and the files it refers to together.

Downloading that folder into several environments creates several snapshots. It does not establish a connection that updates them. For each retained export, record:

- Which brand version it contains.
- Where it is installed.
- How you will replace it after a change.

Keep the folder structure intact so its relative links work. If the tool can later retrieve task context directly, you can remove the export from that workflow.

## Check a change across the connected workflows

Once the assistants are connected, check that each workflow receives a brand update.

Use a genuine brand update when one is needed. Rainbrand [prepares an edit for inspection before it is applied](https://docs.rainbrand.com/guides/edit-brand). The authorised caller can inspect, refine, apply or discard it. After applying the change, request a fresh Brief in each affected workflow.

Check both the returned guidance and the work made from it. If the guidance is current but the output uses an old logo, inspect the files handed to the producing tool. If the guidance itself is old, check the selected brand, reused Brief and any saved snapshot.

A completed Brief stays tied to the version it was prepared from. An export stays a snapshot. Applying a change does not rewrite either one or edit settings inside another assistant. The [guide to updating AI tools](https://www.rainbrand.com/blog/update-brand-guidelines-across-ai-tools/) covers replacing those remaining copies.

We have not run a multi-assistant trial for this article. The procedure is based on the documented workflow and the downloadable example. Different models can produce different outputs from the same brand guidance.

## Use the same reference inside your own product

An embedded assistant needs the same arrangement. Your application identifies the customer's brand, retrieves task context and passes it to every step making the work. Rainbrand's [embedded-product guide](https://docs.rainbrand.com/getting-started/embed-into-your-product) covers connecting that workflow through its API.

Keep the Brand ID and returned version with the output record, so a stale result can be traced to the context used. [Runtime retrieval](https://www.rainbrand.com/blog/how-agents-retrieve-brand-context-at-runtime/) explains the request, caching and failure handling.

Start with two workflows your team already uses. Connect them to the same brand, remove their old saved guidance and inspect the context returned for each task. Record any workflow still using an export in the same inventory.

## More in AI tools

- 28 Oct 2026 6 min read

    ### [How to keep AI-generated content on brand](https://www.rainbrand.com/blog/keep-ai-generated-content-on-brand/)

    Keep agents working from current brand guidance, give each task the right rules and files, and use checks of the finished work to improve the next task.

- 28 Oct 2026 7 min read

    ### [Which MCP servers provide brand guidelines and assets?](https://www.rainbrand.com/blog/mcp-servers-for-brand-guidelines/)

    Brand MCP servers differ in whose brand they cover and what an agent gets back. From each server’s documentation, here is which of Rainbrand, Frontify, Bynder, Standards, Canva, Figma, Brandfetch and Pulumi’s own brand server fits which job.

- 28 Oct 2026 7 min read

    ### [How to update brand guidelines across all your AI tools](https://www.rainbrand.com/blog/update-brand-guidelines-across-ai-tools/)

    Apply a brand change, refresh the context used by work in progress, and replace uploaded or exported copies. Use a worksheet to track which AI workflows have picked it up.

One brand, every agent.

[Connect your agents](https://docs.rainbrand.com/mcp/getting-started)