Skip to content
AI Role Factory
Start
Help and support

Get help with your AI role, or ask us to look after it for you

Start with the outcome you need. We can help your team use the role, fix a problem, improve the setup, or take care of ongoing maintenance.

What do you need?

Choose the quickest route to useful help

Help using the role

Get support with a routine, a connected tool, or helping the team use the role well.

Discuss support →

Improve or maintain it

Ask us to review the setup, make safe changes, and keep it useful as the business evolves.

See ongoing care →

I’m looking after it myself

Open the optional technical guide for health checks, ownership rules, and safe update previews.

Open the technical guide →

Straight answers

Common questions about AI Role Factory

What is AI Role Factory?

AI Role Factory creates a workspace for a familiar business role, such as Chief of Staff, Head of Operations, or Fractional CTO. The workspace gives your chosen AI tool a clear job, useful routines, business context, and boundaries.

Is it another platform my team has to use?

No. The website helps create and deliver the role. Day-to-day work happens through supported AI and business tools your team already knows, and AI Role Factory is not required as a central runtime after delivery.

Who owns the role and its business information?

Your business controls its copy of the role, where it is stored, which accounts are connected, and which changes are accepted. Credentials stay in the selected host or, when an integration needs a credential file, in a private secrets file inside the customer-controlled workspace. The file is excluded from version control and should not be shared.

Can you install and maintain it for us?

Yes. You can build a starter role yourself, ask for help choosing and installing one, or use an ongoing care service. Installation and care do not transfer ownership of the workspace or connected accounts.

How is AI Role Factory different from Legate?

AI Role Factory gives you a role workspace that your business can own and use. Legate is a separate operated service in which an accountable operator runs a defined workflow and reports the outcome. It does not inherit this site’s wizard, customer data, connected accounts, or runtime.

Technical maintenance guide

Quick start

Check first. Preview second.

Run these from any directory. Replace the example paths with your installed workspace and the fresh target workspace you want to compare.

1. Check workspace health

Finds missing files, unresolved setup values, invalid configuration, instruction drift, and unsafe secret placement.

Doctor
npx @ai-role-factory/cli@preview doctor ./my-workspace

2. Preview an update

Compares your installed workspace with a newly generated target. It reports what could change, but applies nothing.

Dry-run update
npx @ai-role-factory/cli@preview plan-update ./my-workspace \
  --target ./new-version

Where the workspace lives

Delivered here. Used and stored there.

The hosted builder and CLI create the same kind of workspace. After generation, client-controlled storage and the selected AI host become the working environment.

1

Deliver

Generate with the browser or CLI and download the workspace files.

2

Store

Keep the workspace in a client folder, private repository, or approved file store.

3

Use

Open it in the supported AI host selected during setup and connect business tools through approved authentication.

See the full ownership and distribution model →

Your instance manifest

The workspace knows where it came from

Look for .ai-role-factory/instance.json. It records the installed release and a checksum and ownership rule for each generated path. It contains no file contents, personal details, credentials, or customer data.

Factory-owned

Role definitions, routines, prompts, skills, and operations guidance.

The factory can propose a replacement only while the installed file still matches its original baseline. Your edit becomes a conflict.

Customer-owned

Your profile, operating scope, session notes, MCP config, environment files, and secrets.

Always preserved. New files you add are preserved too.

Mixed

AGENTS.md, CLAUDE.md, README.md, .gitignore, and integration notes.

Customer-only edits are preserved. If both you and the factory changed the file, the plan reports a conflict for you to review.

Safe update flow

A plan you review, not an update you hope survived

The current release deliberately stops before writing. It gives you enough evidence to make the change yourself or supervise another tool.

  1. 01

    Back up the installed workspace

    Commit it to a private repository or copy the directory. Back up secrets separately in an approved encrypted store.

  2. 02

    Generate a fresh target

    Use the same roles, modules, and intake settings. Web users can run the builder again. CLI users can reuse their saved config file.

  3. 03

    Run doctor

    Fix objective errors first. Keep expected warnings about your mixed-file edits in view.

  4. 04

    Run plan-update

    Review every preserve and conflict result. Move only the approved changes into your installed workspace.

Generate a target with saved config
npx @ai-role-factory/cli@preview \
  --config ./my-company.yaml \
  --output ./new-version
Save the plan as JSON
npx @ai-role-factory/cli@preview plan-update ./my-workspace \
  --target ./new-version --json > update-plan.json

How to read the plan

add
A new factory file is available.
update
The factory changed a file that you have not edited.
preserve
Your file or local edit stays as it is.
remove
The factory removed a generated file that you have not edited.
conflict
Applying the factory change could overwrite your work. Review it manually.

Common technical questions

Do I need to keep using the AI Role Factory website?

No. The website delivers the workspace. After download, the workspace can remain in your own folder or private repository and run through the supported AI host you selected.

Where should the workspace live?

Use a client-controlled folder, private repository, or approved file store. Keep credentials in the host or an approved secrets store. If an integration uses a private secrets file inside the workspace, keep that file out of version control and never share it.

Does AI Role Factory see my workspace?

The factory does not see the contents of your workspace after you download it. The email address you submit is used to deliver the private builder link. When you generate a workspace, the setup and contact details you entered are sent for relevant product and service follow-up. The factory has no default access to the prompts, memory, credentials, or business data you later add. The manifest and maintenance commands run locally and send no telemetry.

Will plan-update change any files?

No. It reads the installed workspace and a separately generated target, then prints a plan. There is no automatic apply command in the current release.

What is stored in the instance manifest?

The factory release, manifest schema, installed roles and modules, maintenance commands, file ownership rules, and content checksums. It does not store file contents, identity fields, credentials, or customer data.

What do I need to run these commands?

Node.js 24. The first npx run may download the preview CLI from npm. The checks themselves run locally against the folders you name.

Is a warning the same as an error?

No. Fix objective errors before planning an update. A warning about a mixed file can be expected when you have added your own instructions or notes.

What should I do with a conflict?

Keep your installed file unchanged. Compare it with the target, decide which factory changes you want, and merge those changes manually. The tool never chooses a winner.

Want us to take a look?

We can help fix a problem, improve an existing role, or put a simple care plan in place.

See ongoing care