How to Set Up a Persona in Grok Bot, OpenAI Dots, Claude and Other AI Agents

Updated: November 12, 2026 · By Brian Arrington (Founder, AI Consultant and Business Coach)*

The short answer

Load one persona per agent or Bot. Copy the Persona / System Prompt section from the Markdown file, paste it into your agent's instructions, connect only the listed tools, and send the Example First Task. The setup guide at botplaybooks.com/setup covers Grok Bot, Grok, OpenAI Dots, Claude, Hermes and OpenClaw.

A persona is only useful once it is loaded into the agent you already use. The good news is that it takes a few minutes. The steps below follow the BotPlaybooks setup guide at botplaybooks.com/setup, which covers computers, iPhone, iPad and Android.

Step 1: Get the part you will paste

Each persona is one plain Markdown (.md) file. Packs come as a .zip file with a folder of persona files inside, so unzip it first.

Open the persona file in any text editor or notes app. Find the heading Persona / System Prompt and copy everything under it, down to the next heading, Tools and Connectors. Keep the file open, because you will also need the tools list and the Example First Task.

Step 2: Load it into your agent

The rule of thumb for every platform: one persona per agent or Bot.

Grok Bot

• In your main Grok Bot, use the plus button to give it the persona's Markdown file.
• Tell it to take that file and build the persona as a new agent.
• A Chief of Staff or CEO bot helps. If you have one, tell it that it's in charge of the new persona and that the new persona reports to it.
• Then send the Example First Task.

Grok

• Open the place where Grok lets you give it standing instructions (its customize or instructions settings), paste the Persona / System Prompt section and save, then start a new chat.
• If you would rather not change your settings, paste the section as your first message in a new chat and ask Grok to follow it for the rest of the chat.

OpenAI Dots

• Open your dot and give it the Persona / System Prompt section as its working instructions. Turn the guardrails into Custom Rules. Full steps are on the OpenAI Dots guide at botplaybooks.com/#dots and the setup guide at botplaybooks.com/setup.

Claude

• Go to claude.ai/projects, click New Project and give it the persona's name.
• Choose Set project instructions, paste the Persona / System Prompt section and save.
• Optional: add the .md file to the project's knowledge with the + button.

Hermes and OpenClaw

Per the BotPlaybooks setup guide, both read their identity from a file called SOUL.md. Locations can vary by version and setup, so check each project's own docs. For Hermes it is usually in your Hermes home folder (\~/.hermes/SOUL.md). For OpenClaw it is in your agent workspace (default \~/.openclaw/workspace), and you can also edit it in the Control UI under Settings, Agents, Files. In both cases, copy your current SOUL.md somewhere safe first, paste in the Persona / System Prompt section, save, and start a new session so the agent loads it.

If you use a different tool, email [email protected] and tell us which one. We'll walk you through how to load your persona.

Step 3: Connect only what the persona needs, then test

Connect only the tools listed under Tools and Connectors. Fewer connections mean less that can go wrong. Then send the persona's Example First Task as your first message. It is the fastest way to see what the persona does with your real information.

Review everything before you act on it. Your agent drafts; you approve anything sent, posted or paid.

Stuck?

The setup guide includes answers to common questions, like where a downloaded file went and how to open a .zip on each device. If you are stuck on a step, BotPlaybooks support is at [email protected]. Tell them which step and which agent you are using.

Which agent will you load your first persona into?

BotPlaybooks is independent and not affiliated with any agent platform. Product names belong to their owners.

BotPlaybooks is a brand of Arrington Consulting LLC.

Disclaimer: This post is general information, not professional, legal or financial advice. Results vary. AI can make mistakes, so always review outputs before acting.