User testing your manual with Megapersonas a new AI technique for technical writers

There’s a new prompting technique in town: megapersonas. Instead of asking ChatGPT to role-play a single expert, you now simulate hundreds or even thousands of realistic users, instantly. Thanks to large persona datasets like FinePersonas and PersonaHub, you can now run massive user research simulations from your laptop.

This means you can test your documentation with entire user segments before it ever reaches a real reader.

Why it matters for technical writers

User testing is often time-consuming and expensive. But megapersonas let you simulate targeted feedback on your manual from dozens or even hundreds of realistic personas:

• Beginners vs. advanced users
• Different industries and job roles
• Varying technical skills or accessibility needs

You can test clarity, terminology, order of instructions and more, without having to organise a single Zoom call.

How to apply this in your writing workflow

Step 1: Define your test goals

What do you want to learn from your simulated users?

• Is your manual easy to follow?
• Are the steps in the right order?
• Are technical terms too advanced?
• Is anything missing or unclear?

Step 2: Use this prompt to simulate feedback

Use the FinePersonas database to simulate 250 personas with the following characteristics:

• Varying experience levels with technical documentation (beginner to advanced)
• Roles such as system administrator, help desk agent, or end user
• Sectors: IT, healthcare, and government

Ask them to review the manual and provide feedback on:

• Clarity of the instructions
• Sequence and logic of the steps
• Use of technical terms (too complex or appropriate?)
• Missing or confusing information

Present the feedback as grouped user comments, including concrete suggestions for improvement.

Follow-up prompt:

“Summarise the key themes from the feedback and highlight the most frequent or critical issues.”

Step 3: Rerun with variations

Change the audience:

• Simulate visually impaired users
• Simulate non-native English speakers
• Simulate users with low digital literacy
• Simulate industry-specific roles (e.g., lab technicians, warehouse workers)

Step 4: Use the results to improve your manual

Update the content and rerun the test with a fresh set of personas. This lets you refine your writing with each cycle.

What is the FinePersonas database?

FinePersonas is an open-source dataset of over 21 million detailed personas, designed to help simulate realistic user behaviour and feedback through AI.

Each persona includes characteristics like:

• Age, gender, location
• Job and education level
• Interests, technical skill, and even communication style

You can use these personas in your prompts without loading any files. The model draws on learned patterns and will generate realistic simulations.

Available for free via Hugging Face:
FinePersonas on Hugging Face

Prompt Examples You Can Copy

Quick usability check:

“Simulate 100 non-technical users from different industries. Have them read my setup guide and describe the first step they found confusing.”

For accessibility testing:

“Simulate 50 users with visual impairments. Ask them to evaluate whether the manual is understandable without relying on images.”

For UX feedback from support staff:

“Simulate 25 help desk agents who often assist customers using our tool. Let them assess if the manual helps reduce support calls.”

Final thoughts

Megapersonas aren’t a replacement for real users, but they are a powerful tool for rapid iteration and early validation. They help uncover blind spots in your documentation before rollout and they scale like nothing else.

In other words, they turn ChatGPT into your own personal UX research team.

Give it a try in your next review cycle. You might be surprised at what your virtual audience has to say.