AI needs enterprise context. Merge’s new Knowledge Base Unified API delivers it

We’re thrilled to announce the launch of Merge’s newest Unified API: Knowledge Base. Starting today, we’re opening up access to Atlassian Confluence, with even more exciting additions, like Notion and Salesforce Knowledge, on the horizon.

Knowledge Base is built to help teams connect knowledge data for AI use cases at an enterprise scale. We couldn’t be more excited to see what you build.

From cave walls to AI agents

Human history is the story of how we preserve knowledge. We first carved what we knew into stone, bound it into books, and then stored it across silicon circuits.

Trained on the entire corpus of human knowledge, LLMs from OpenAI, Anthropic, and Mistral can answer almost any question a person has in seconds.

Now, we’re entering a new phase.

While AI agents can now read, reason, and respond at lightning speed, they’re only as powerful as the data they have access to. To truly deliver value to enterprises, AI needs access to proprietary knowledge stored within an organization’s tech stack.

To get that access? AI needs connections that work at scale. AI needs integrations. AI needs Merge.

Related: A guide to knowledge base integrations

No one wants to build Knowledge Base integrations

At first glance, Knowledge Base seems like a simple category. Unlike other categories, which can have over 50 integrations, there are only a handful of major tools. But once we started digging into the challenges faced by top AI labs, enterprises, and fast-growing startups who all wanted knowledge data, it became clear: this category is a serious technical challenge.

• Non-hierarchical data models: AI use cases require normalized data. Knowledge Base as a category is anything but normal

For example, while Confluence offers a basic page hierarchy within Spaces, the reality is far messier. Pages can be deeply nested, manually cross-linked, or frequently restructured, creating complex relationships. Turning this web into a normalized parent/child model requires careful untangling. You can’t afford to misrepresent relationships, but you must simplify them enough for downstream systems to make sense of the data.

• High stakes permissions: Integrations in this category must respect access control lists (ACLs): for example, ensuring a sales rep can’t access a manager’s private performance review notes in a restricted space. This is a challenge in a platform like Notion, Confluence, or Salesforce Knowledge: different folders and pages can have differing permissions around access, even if they are in the same hierarchy

• Syncing at scale: The scale of knowledge base data is astounding. A company using Salesforce Knowledge will have thousands of pages of variable length. And while some users want all their data, others only want certain articles or workspaces

• Rate limits - Rate limits are a challenge for any integration, but Confluence rate limits provide a unique challenge. Unlike traditional limit ceilings which have you retry after n seconds, Confluence uses an adaptive system that makes it difficult to predict when you'll hit a limit

And that’s all on top of the usual integration headaches one encounters when building integrations themselves.

How the Merge Knowledge Base Unified API helps

Knowledge Base isn’t just another integration category. It’s the foundation of Enterprise AI. 

Merge’s Knowledge Base Unified API gives teams everything they need to bring internal knowledge into AI systems, quickly, accurately, and securely

• Unified data model: We’ve created six new Common Models that make knowledge access simple and consistent. Confluence’s deeply nested spaces and pages—and Notion’s block-based structure—fit cleanly into our Containers and Articles models, making it easy for developers to access and embed knowledge data at scale

• Built-in permission handling: We’ve made ACLs something you don’t have to worry about. Confluence’s space- and page-level permissions are mapped into a permissions model, ensuring users only access the content they’re authorized to see

• Up-to-date data: We’ve continued to expand our best-in-class infrastructure to ensure data stays fresh without overwhelming APIs. With features like Selective Sync and webhooks, customers get current Confluence content on demand without needing to sync entire workspaces

Merge doesn’t just help you integrate faster: we help you win.

Article Picker makes your customers’ lives easier

Not every team wants to sync their entire knowledge base. With Merge, they don’t have to.

Article Picker lets your users choose exactly which spaces, folders, or pages to bring into your product at the time of data sync. Whether it's a Head of Security setting up enterprise search, or a GTM lead preparing an enablement lesson by only syncing content from a specific folder, Article Picker makes it easy to tailor what gets synced.

Even better, Article Picker is built directly into the Merge Link flow and works out of the box. Your users get full precision, without complexity.

Available now for Confluence

Merge’s Knowledge Base Unified API is live for Atlassian Confluence in open beta. And we’re not stopping there, we’re building integrations for:

• Salesforce Knowledge
• Zendesk Help Center
• Notion
• Intercom Help Center
• Other systems your customers rely on every day

Ready to see it in action? Get in touch, or check out the Knowledge Base docs.

Let’s help your product connect to your customers’ knowledge.