AI Features
We offer a range of AI features designed to streamline your tasks and enhance efficiency.
By default, all features are disabled and require activation by an admin. You will find them under Project Settings → AI → AI Agents & MCP, split across an Editor tab for the features your team uses and a Readers tab for what your readers' AI tools can do with your documentation.
The features are:
AI Agent: Ask for a change in a conversation and have it drafted across your pages, API references and changelog posts for you to review.
Self-Updating Docs: Attach your code repositories so the agent can check what your docs claim against what your code does, and draft the updates a pull request implies.
AI Writing Tools: Select text in the editor, pick a function, and have it rewritten in place. Includes shortening, enhancing and grammar correction.
AI Assistant: Ask questions about the docs in natural language and receive GPT powered answers.
AI Commit Messages: Automatically annotate page histories.
AI SEO Helper: Summarises pages to write a META description.
MCP Servers: Connect AI applications with your docs, so readers can search them and editors can write them.
WebMCP: Let AI agents in your readers' browsers search, read and open your pages.
Feedback Spam Filter: Filters spam in feedback messages automatically.
Redact PII from Feedback: Redacts personal identifiable information from feedback messages automatically.
Read the linked sections to understand what information we send for each feature.
AI Credits
AI credits pay for AI Agent, including the pull request checks under Self-Updating Docs. No other feature on this page spends them.
A plan with AI features includes a monthly allowance of credits, which renews with your billing period. How many depends on the plan; see Pricing. Credits belong to a project, so each of your projects has its own allowance.
Each message you send the agent spends credits from the balance. How many depends on the size of the job, so correcting a page costs less than restructuring a section.
You can see what a project has left in three places:
The AI Editor window, in its top bar.
This page, at the end of the row holding the Editor and Readers tabs.
Project Settings → Billing → Plan & Usage, under AI credits, which shows the full picture and when the allowance renews.
Admins are emailed once when a project drops to 20% of its allowance. You can turn that email off in Account Settings.
Running out
With no credits left, the agent will not start a new run, and pull request checks stop until the allowance renews or you top up. A run already under way finishes what it can and keeps everything it has staged for you to review.
The AI Editor window says so above its message box, along with when your credits renew. Admins can select Get credits there to top up.
Everything else carries on as normal, including your published documentation and every other AI feature.
Topping up
To buy more before your allowance renews, open Project Settings → Billing → Plan & Usage and use Top up under AI credits. Bought credits sit on top of your monthly allowance, survive a plan change, and last 12 months. Your allowance is always spent first.
On an enterprise plan, credits are a term of your contract instead. Contact Us and we will size them to your team.
Which models we use
The documentation agent, and the pull request checks that run on it, use the model an admin picks for the project under Project Settings → AI → AI Agents & MCP → Editor → Agent model. Each option shows the provider that serves it and the region it runs in, so you can rule out a jurisdiction if you need to, and Auto follows our current recommendation.
Those runs are served through OpenRouter, and only providers with zero data retention are used. Your content is never kept and never used to train a model.
Our other AI features, including AI Writing Tools, AI commit messages, AI Assistant and the feedback filters, use OpenAI's GPT models. AI Writing Tools are not affected by the model an admin picks: rewriting a selection always runs on a small fast model.
Terms & Conditions
For the features that use OpenAI, quoting from OpenAI's Data Controls:
"Your data is your data. As of March 1, 2023, data sent to the OpenAI API is not used to train or improve OpenAI models".
However:
"OpenAI retains API data for 30 days for abuse and misuse monitoring purposes. A limited number of authorized OpenAI employees, as well as specialized third-party contractors that are subject to confidentiality and security obligations, can access this data solely to investigate and verify suspected abuse."
By using any AI services or features that DeveloperHub provides, you accept the transfer of the relevant data to the provider serving that feature.
Need help? Visit our community forums or contact us.