Go from generic docs to tailored experiences and AI assistance

Go from generic docs to tailored experiences and AI assistance

Product updates

Product updates

Product updates

31 Jul, 2025

Author

Author

Author

The GitBook logo with a start overlayed, inside a white square box on a pink, blue and white background
The GitBook logo with a start overlayed, inside a white square box on a pink, blue and white background
The GitBook logo with a start overlayed, inside a white square box on a pink, blue and white background


We’ve been talking a lot recently about the future of documentation.

Today, documentation is one-size-fits-all and disconnected from your product. Instead, we think docs should adapt to each user and integrate more closely with your product — delivering a tailored, intelligent experience that evolves alongside your users.

Today we’re excited about a major launch that make this vision a reality — adaptive content, with an all-new GitBook Assistant.

Today we’re excited to launch our first features that make it a reality — adaptive content and an all-new AI assistant.

Highlights

  • Adaptive content is here – Give every individual user a tailored, targeted docs experience using cookies, URL strings and more. Check out our live demo!

  • A new AI-powered Assistant – We’ve reimagined AI search to bring powerful new knowledge discovery to your site visitors through chat.

  • GitBook Assistant integrates with adaptive content – Providing personalized answers with user-aware context.

  • Connect MCP servers to Assistant – Pull information from third-party sources and trigger actions like changing settings or filing support tickets.

From paper maps to GPS-powered docs

Documentation hasn’t changed much in the last few decades. It may look sleeker, but it’s still a static, identical experience — a paper map showing every possible route, destination and stopping point, with no awareness of who’s reading it.

But no two users are the same.

We believe that needs to change. Modern docs need to be more dynamic, and better optimized for individual users. They need to be more like a GPS.

An illustration of a GPS style navigation system. The driver avatar has a GitBook logo and along the route are stops labelled 'Developer program', 'Troubleshooting', 'Advanced configuration' and 'API Reference'. It all sits  on a pink, blue and white background

New users need onboarding. Experienced developers want API reference docs. Prospective customers want overviews and USPs that make your product stand out. Each deserve a tailored route — optimized from their starting point to their goal, with the right stops points along the way.

This is what we’re building: documentation that adapts to each visitor.

Not a map. A GPS.

And it’s available right now in GitBook.

Adaptive content — docs that respond to your users

The first part of this big picture is adaptive content, which changes your docs depending on who is viewing them.

Adaptive content builds a user profile using data from different sources — including your product, the user themselves, and external sources. And you can use this data to choose which content each user sees.

That means you can create different landing pages optimized for users depending on where they are in the product journey:

  • Got a new user who’s onboarding? They’ll see a ‘Getting Started’ page with setup guides.

  • Prospective customers? Show them a landing page with product highlights and links to pages covering your USPs.

  • Established technical user? Skip the basics and show them the API docs with pre-filled keys, advanced guides, and more.

Even simple setups can dramatically improve the user experience. Want to see how it works? Visit our demo site and explore different user journeys in real time.

What can adaptive content do?

Adaptive content scales from simple to sophisticated.

Start with basic personalization, like pre-filled API keys, different header buttons for logged in users, or contextual pages based on cookies or URL parameters.

Then, as you start bringing in more user data, you can build more complex ‘claims’ — the logic that defines how content should respond to specific traits or behaviors.

That means you can tailor docs for customer vs non-customer journeys — one designed to convert, the other to offer incredible support. Or recommend features to specific user groups to encourage upgrades.

And there are a few different ways to pass visitor data to GitBook once your claims are up and running. You can show different content based on cookies, URL query parameters, a feature flag provider, or authenticated access.

A screenshot of the GitBook scheme builder  on a pink, blue and white background

Right now, you can use adaptive content to show and hide:

  • Pages, site sections, and variants

  • Header buttons or links

  • Specific blocks on a page

  • Dynamic variables in inline expressions (e.g. user names, account types etc)

And of course, adaptive content can also auto-fill API keys in API endpoint blocks.

Setting up adaptive content

Want to try adaptive content for yourself? It’s live now for Ultimate sites — sign in or sign up for free to take it for a spin.

Head over to our docs to find out how to get started and how to start adding data and setting up claims. Alternatively, follow along with our guide to set up your first piece of adaptive content via cookies.

GitBook Assistant — knowledge from your docs and beyond

As part of the adaptive content launch, we’re also introducing GitBook Assistant — a powerful new AI experience for your docs. 

It works with adaptive content in powerful ways — but more on that in a moment.

Assistant is a big step up over our previous AI search functionality. While our old solution was fast and gave end-users accurate answers based on your docs, GitBook Assistant offers a new chat-based UI, seamless integration with adaptive content, and the option to connect with MCP servers to provide better answers with more context.

Smarter AI, deeper understanding

Our previous version of AI search used retrieval-augmented generation (RAG). It worked by pulling keywords from the user’s question, choosing pages based on those keywords, then feeding the LLM information from those pages to answer the question.

It worked pretty well, but was limited to the content of your docs, with no option for follow-up questions

GitBook Assistant changes that.

First, it uses agentic retrieval, giving it a deeper understanding of user intent, and more accurate responses — even to complex questions.

Secondly, it’s integrated with adaptive content, so answers can reflect the user’s profile. For example, free-tier users asking about a paid feature will get a clear, helpful response — but Assistant can also tell them they don’t have access, and include a link to upgrade.

But GitBook Assistant has another trick up its sleeve — it can connect to your product and other sources using MCP servers.

And that unlocks powerful possibilities.

Answers and actions that go beyond your docs

By connecting to MCP servers, GitBook Assistant can pull information from different sources and use that information answer questions with even more context.

But it can also send commands back to those servers, allowing your users to carry out actions right from the GitBook Assistant chat window.

An illustration of different product icons — including LaunchDarkly, Slack, Bucket, GitHub, Intercom and Linear — flowing into the GitBook Assistant icon in the center. This all sits  on a pink, blue and white background

Here are a couple of quick examples:

  • Connect Assistant to your GitHub Community and it can answer questions about previous discussions.

  • Link to tools like Intercom or Linear to file support tickets or report a bug — right from the chat window.

With GitBook Assistant, your documentation becomes an interface — not just a resource.

From static docs to intelligent experiences

Adaptive content, and its deep integration with GitBook Assistant, marks a turning point in how docs are built — and what they are capable of.

It helps you deliver the right content to the right person, at the right time.

And at the same time, GitBook Assistant makes your docs more interactive, more intelligent, and more actionable.

This launch transforms documentation from a static map into a GPS — guiding each user to success in real time.

Ready to explore?


We’ve been talking a lot recently about the future of documentation.

Today, documentation is one-size-fits-all and disconnected from your product. Instead, we think docs should adapt to each user and integrate more closely with your product — delivering a tailored, intelligent experience that evolves alongside your users.

Today we’re excited about a major launch that make this vision a reality — adaptive content, with an all-new GitBook Assistant.

Today we’re excited to launch our first features that make it a reality — adaptive content and an all-new AI assistant.

Highlights

  • Adaptive content is here – Give every individual user a tailored, targeted docs experience using cookies, URL strings and more. Check out our live demo!

  • A new AI-powered Assistant – We’ve reimagined AI search to bring powerful new knowledge discovery to your site visitors through chat.

  • GitBook Assistant integrates with adaptive content – Providing personalized answers with user-aware context.

  • Connect MCP servers to Assistant – Pull information from third-party sources and trigger actions like changing settings or filing support tickets.

From paper maps to GPS-powered docs

Documentation hasn’t changed much in the last few decades. It may look sleeker, but it’s still a static, identical experience — a paper map showing every possible route, destination and stopping point, with no awareness of who’s reading it.

But no two users are the same.

We believe that needs to change. Modern docs need to be more dynamic, and better optimized for individual users. They need to be more like a GPS.

An illustration of a GPS style navigation system. The driver avatar has a GitBook logo and along the route are stops labelled 'Developer program', 'Troubleshooting', 'Advanced configuration' and 'API Reference'. It all sits  on a pink, blue and white background

New users need onboarding. Experienced developers want API reference docs. Prospective customers want overviews and USPs that make your product stand out. Each deserve a tailored route — optimized from their starting point to their goal, with the right stops points along the way.

This is what we’re building: documentation that adapts to each visitor.

Not a map. A GPS.

And it’s available right now in GitBook.

Adaptive content — docs that respond to your users

The first part of this big picture is adaptive content, which changes your docs depending on who is viewing them.

Adaptive content builds a user profile using data from different sources — including your product, the user themselves, and external sources. And you can use this data to choose which content each user sees.

That means you can create different landing pages optimized for users depending on where they are in the product journey:

  • Got a new user who’s onboarding? They’ll see a ‘Getting Started’ page with setup guides.

  • Prospective customers? Show them a landing page with product highlights and links to pages covering your USPs.

  • Established technical user? Skip the basics and show them the API docs with pre-filled keys, advanced guides, and more.

Even simple setups can dramatically improve the user experience. Want to see how it works? Visit our demo site and explore different user journeys in real time.

What can adaptive content do?

Adaptive content scales from simple to sophisticated.

Start with basic personalization, like pre-filled API keys, different header buttons for logged in users, or contextual pages based on cookies or URL parameters.

Then, as you start bringing in more user data, you can build more complex ‘claims’ — the logic that defines how content should respond to specific traits or behaviors.

That means you can tailor docs for customer vs non-customer journeys — one designed to convert, the other to offer incredible support. Or recommend features to specific user groups to encourage upgrades.

And there are a few different ways to pass visitor data to GitBook once your claims are up and running. You can show different content based on cookies, URL query parameters, a feature flag provider, or authenticated access.

A screenshot of the GitBook scheme builder  on a pink, blue and white background

Right now, you can use adaptive content to show and hide:

  • Pages, site sections, and variants

  • Header buttons or links

  • Specific blocks on a page

  • Dynamic variables in inline expressions (e.g. user names, account types etc)

And of course, adaptive content can also auto-fill API keys in API endpoint blocks.

Setting up adaptive content

Want to try adaptive content for yourself? It’s live now for Ultimate sites — sign in or sign up for free to take it for a spin.

Head over to our docs to find out how to get started and how to start adding data and setting up claims. Alternatively, follow along with our guide to set up your first piece of adaptive content via cookies.

GitBook Assistant — knowledge from your docs and beyond

As part of the adaptive content launch, we’re also introducing GitBook Assistant — a powerful new AI experience for your docs. 

It works with adaptive content in powerful ways — but more on that in a moment.

Assistant is a big step up over our previous AI search functionality. While our old solution was fast and gave end-users accurate answers based on your docs, GitBook Assistant offers a new chat-based UI, seamless integration with adaptive content, and the option to connect with MCP servers to provide better answers with more context.

Smarter AI, deeper understanding

Our previous version of AI search used retrieval-augmented generation (RAG). It worked by pulling keywords from the user’s question, choosing pages based on those keywords, then feeding the LLM information from those pages to answer the question.

It worked pretty well, but was limited to the content of your docs, with no option for follow-up questions

GitBook Assistant changes that.

First, it uses agentic retrieval, giving it a deeper understanding of user intent, and more accurate responses — even to complex questions.

Secondly, it’s integrated with adaptive content, so answers can reflect the user’s profile. For example, free-tier users asking about a paid feature will get a clear, helpful response — but Assistant can also tell them they don’t have access, and include a link to upgrade.

But GitBook Assistant has another trick up its sleeve — it can connect to your product and other sources using MCP servers.

And that unlocks powerful possibilities.

Answers and actions that go beyond your docs

By connecting to MCP servers, GitBook Assistant can pull information from different sources and use that information answer questions with even more context.

But it can also send commands back to those servers, allowing your users to carry out actions right from the GitBook Assistant chat window.

An illustration of different product icons — including LaunchDarkly, Slack, Bucket, GitHub, Intercom and Linear — flowing into the GitBook Assistant icon in the center. This all sits  on a pink, blue and white background

Here are a couple of quick examples:

  • Connect Assistant to your GitHub Community and it can answer questions about previous discussions.

  • Link to tools like Intercom or Linear to file support tickets or report a bug — right from the chat window.

With GitBook Assistant, your documentation becomes an interface — not just a resource.

From static docs to intelligent experiences

Adaptive content, and its deep integration with GitBook Assistant, marks a turning point in how docs are built — and what they are capable of.

It helps you deliver the right content to the right person, at the right time.

And at the same time, GitBook Assistant makes your docs more interactive, more intelligent, and more actionable.

This launch transforms documentation from a static map into a GPS — guiding each user to success in real time.

Ready to explore?


We’ve been talking a lot recently about the future of documentation.

Today, documentation is one-size-fits-all and disconnected from your product. Instead, we think docs should adapt to each user and integrate more closely with your product — delivering a tailored, intelligent experience that evolves alongside your users.

Today we’re excited about a major launch that make this vision a reality — adaptive content, with an all-new GitBook Assistant.

Today we’re excited to launch our first features that make it a reality — adaptive content and an all-new AI assistant.

Highlights

  • Adaptive content is here – Give every individual user a tailored, targeted docs experience using cookies, URL strings and more. Check out our live demo!

  • A new AI-powered Assistant – We’ve reimagined AI search to bring powerful new knowledge discovery to your site visitors through chat.

  • GitBook Assistant integrates with adaptive content – Providing personalized answers with user-aware context.

  • Connect MCP servers to Assistant – Pull information from third-party sources and trigger actions like changing settings or filing support tickets.

From paper maps to GPS-powered docs

Documentation hasn’t changed much in the last few decades. It may look sleeker, but it’s still a static, identical experience — a paper map showing every possible route, destination and stopping point, with no awareness of who’s reading it.

But no two users are the same.

We believe that needs to change. Modern docs need to be more dynamic, and better optimized for individual users. They need to be more like a GPS.

An illustration of a GPS style navigation system. The driver avatar has a GitBook logo and along the route are stops labelled 'Developer program', 'Troubleshooting', 'Advanced configuration' and 'API Reference'. It all sits  on a pink, blue and white background

New users need onboarding. Experienced developers want API reference docs. Prospective customers want overviews and USPs that make your product stand out. Each deserve a tailored route — optimized from their starting point to their goal, with the right stops points along the way.

This is what we’re building: documentation that adapts to each visitor.

Not a map. A GPS.

And it’s available right now in GitBook.

Adaptive content — docs that respond to your users

The first part of this big picture is adaptive content, which changes your docs depending on who is viewing them.

Adaptive content builds a user profile using data from different sources — including your product, the user themselves, and external sources. And you can use this data to choose which content each user sees.

That means you can create different landing pages optimized for users depending on where they are in the product journey:

  • Got a new user who’s onboarding? They’ll see a ‘Getting Started’ page with setup guides.

  • Prospective customers? Show them a landing page with product highlights and links to pages covering your USPs.

  • Established technical user? Skip the basics and show them the API docs with pre-filled keys, advanced guides, and more.

Even simple setups can dramatically improve the user experience. Want to see how it works? Visit our demo site and explore different user journeys in real time.

What can adaptive content do?

Adaptive content scales from simple to sophisticated.

Start with basic personalization, like pre-filled API keys, different header buttons for logged in users, or contextual pages based on cookies or URL parameters.

Then, as you start bringing in more user data, you can build more complex ‘claims’ — the logic that defines how content should respond to specific traits or behaviors.

That means you can tailor docs for customer vs non-customer journeys — one designed to convert, the other to offer incredible support. Or recommend features to specific user groups to encourage upgrades.

And there are a few different ways to pass visitor data to GitBook once your claims are up and running. You can show different content based on cookies, URL query parameters, a feature flag provider, or authenticated access.

A screenshot of the GitBook scheme builder  on a pink, blue and white background

Right now, you can use adaptive content to show and hide:

  • Pages, site sections, and variants

  • Header buttons or links

  • Specific blocks on a page

  • Dynamic variables in inline expressions (e.g. user names, account types etc)

And of course, adaptive content can also auto-fill API keys in API endpoint blocks.

Setting up adaptive content

Want to try adaptive content for yourself? It’s live now for Ultimate sites — sign in or sign up for free to take it for a spin.

Head over to our docs to find out how to get started and how to start adding data and setting up claims. Alternatively, follow along with our guide to set up your first piece of adaptive content via cookies.

GitBook Assistant — knowledge from your docs and beyond

As part of the adaptive content launch, we’re also introducing GitBook Assistant — a powerful new AI experience for your docs. 

It works with adaptive content in powerful ways — but more on that in a moment.

Assistant is a big step up over our previous AI search functionality. While our old solution was fast and gave end-users accurate answers based on your docs, GitBook Assistant offers a new chat-based UI, seamless integration with adaptive content, and the option to connect with MCP servers to provide better answers with more context.

Smarter AI, deeper understanding

Our previous version of AI search used retrieval-augmented generation (RAG). It worked by pulling keywords from the user’s question, choosing pages based on those keywords, then feeding the LLM information from those pages to answer the question.

It worked pretty well, but was limited to the content of your docs, with no option for follow-up questions

GitBook Assistant changes that.

First, it uses agentic retrieval, giving it a deeper understanding of user intent, and more accurate responses — even to complex questions.

Secondly, it’s integrated with adaptive content, so answers can reflect the user’s profile. For example, free-tier users asking about a paid feature will get a clear, helpful response — but Assistant can also tell them they don’t have access, and include a link to upgrade.

But GitBook Assistant has another trick up its sleeve — it can connect to your product and other sources using MCP servers.

And that unlocks powerful possibilities.

Answers and actions that go beyond your docs

By connecting to MCP servers, GitBook Assistant can pull information from different sources and use that information answer questions with even more context.

But it can also send commands back to those servers, allowing your users to carry out actions right from the GitBook Assistant chat window.

An illustration of different product icons — including LaunchDarkly, Slack, Bucket, GitHub, Intercom and Linear — flowing into the GitBook Assistant icon in the center. This all sits  on a pink, blue and white background

Here are a couple of quick examples:

  • Connect Assistant to your GitHub Community and it can answer questions about previous discussions.

  • Link to tools like Intercom or Linear to file support tickets or report a bug — right from the chat window.

With GitBook Assistant, your documentation becomes an interface — not just a resource.

From static docs to intelligent experiences

Adaptive content, and its deep integration with GitBook Assistant, marks a turning point in how docs are built — and what they are capable of.

It helps you deliver the right content to the right person, at the right time.

And at the same time, GitBook Assistant makes your docs more interactive, more intelligent, and more actionable.

This launch transforms documentation from a static map into a GPS — guiding each user to success in real time.

Ready to explore?

Get the GitBook newsletter

Get the latest product news, useful resources and more in your inbox. 130k+ people read it every month.

Email

Similar posts

Get started for free

Play around with GitBook and set up your docs for free. Add your team and pay when you’re ready.

Trusted by leading technical product teams

The NVIDIA logo
The Carta logo
The Ericsson logo
The Cisco logo
The Fedex logo
The Zoom logo
  • The Braze logo
  • The Bucket logo
  • The Cortex logo
  • The Count logo
  • The Digibee logo
  • The Gravitee logo
  • The Hebbia logo
  • The HockeyStack logo
  • The Ideogram logo
  • The JAM logo
  • The Make logo
  • The Material logo
  • The Multiplier logo
  • The Nightfall AI logo
  • The Onum logo
  • The Photoroom logo
  • The Pylon logo
  • The Relay.app logo
  • The Rox logo
  • The ScraperAPI logo
  • The Seam logo
  • The Sendbird logo
  • The Sola logo
  • The Synk logo
  • The Tabnine logo
  • The ZenML logo

Get started for free

Play around with GitBook and set up your docs for free. Add your team and pay when you’re ready.

Trusted by leading technical product teams

The NVIDIA logo
The Carta logo
The Ericsson logo
The Cisco logo
The Fedex logo
The Zoom logo
  • The Braze logo
  • The Bucket logo
  • The Cortex logo
  • The Count logo
  • The Digibee logo
  • The Gravitee logo
  • The Hebbia logo
  • The HockeyStack logo
  • The Ideogram logo
  • The JAM logo
  • The Make logo
  • The Material logo
  • The Multiplier logo
  • The Nightfall AI logo
  • The Onum logo
  • The Photoroom logo
  • The Pylon logo
  • The Relay.app logo
  • The Rox logo
  • The ScraperAPI logo
  • The Seam logo
  • The Sendbird logo
  • The Sola logo
  • The Synk logo
  • The Tabnine logo
  • The ZenML logo

Get started for free

Play around with GitBook and set up your docs for free. Add your team and pay when you’re ready.

Trusted by leading technical product teams

The NVIDIA logo
The Carta logo
The Ericsson logo
The Cisco logo
The Fedex logo
The Zoom logo
  • The Braze logo
  • The Bucket logo
  • The Cortex logo
  • The Count logo
  • The Digibee logo
  • The Gravitee logo
  • The Hebbia logo
  • The HockeyStack logo
  • The Ideogram logo
  • The JAM logo
  • The Make logo
  • The Material logo
  • The Multiplier logo
  • The Nightfall AI logo
  • The Onum logo
  • The Photoroom logo
  • The Pylon logo
  • The Relay.app logo
  • The Rox logo
  • The ScraperAPI logo
  • The Seam logo
  • The Sendbird logo
  • The Sola logo
  • The Synk logo
  • The Tabnine logo
  • The ZenML logo