For immedio, documentation isn’t just a support asset — it’s how customers successfully use a highly configurable product. immedio is an AI driven lead-to-meeting conversion platform, embedded directly into each customer's website. Every implementation is different — so immedio's docs need to handle that nuance, edge cases, and constant variation without creating friction for support or engineering.

After rebuilding their help center on GitBook, immedio reduced escalations from support to engineering by 30% and saw a noticeable drop in questions as more customers were able to self-serve.

We spoke with engineering manager Takuma Inoue about how immedio turned documentation into a system of record — and how AI helps them keep it accurate as the product evolves.

The challenge: edge cases and scattered answers

immedio’s product behaves differently depending on how each customer integrates it. Small frontend differences can change which settings apply, and many support questions require engineering context to resolve.

Before GitBook, immedio’s knowledge was scattered. Their help articles lived in Zendesk, company knowledge in Slack threads and channels, and one-off notes and ideas in Notion.

“We wanted the help center to function as the single, trustworthy source of information. When knowledge is scattered across Zendesk, Slack, and Notion, nobody knows what to trust.”

Updates were slow, too. One product manager owned most documentation changes, creating a bottleneck that pulled time away from higher-impact work.

Making documentation the system of record

Immedio migrated their help center from Zendesk to GitBook, reducing support conversations and escalations by 30%. But the real change was how documentation fit into daily work.

GitBook let them support two authoring workflows in one place. Engineers could contribute in Markdown through Git-based review, while product managers could edit in the visual editor, without touching Git.

That flexibility removed the PM bottleneck and kept docs closely tied to real product and engineering decisions. Documentation stopped lagging behind reality.

“It wasn’t uncommon for an entire day to disappear into support. The questions were complex and constant — and that made fixing our documentation process critical to growing the product.”

AI-assisted updates — with Git review​

With their docs in Git, immedio could seamlessly build AI-powered workflows into their documentation process. The team uses Devin, an AI agent, to draft documentation updates based on support conversations and engineering context.​

When support escalates a technical question to engineering and it gets resolved, anyone on the team can tag Devin in Slack to update the docs. The AI agent drafts the proposed changes and opens a pull request — ready for a human to review, edit, and merge.

“We never fully hand documentation over to AI. Just like with code, the final quality check has to be done by a human. If your help center says something inaccurate, you lose trust immediately.”

To keep drafts grounded, immedio enforces one rule: any technical claim must be verified against the codebase. That avoids generic explanations and reduces errors before review.

In this workflow, AI provides speed, and humans focus on accuracy.

AI search that feels like a product expert

The team also enabled GitBook Assistant in their docs, changing how both customers and the support team find answers. It uses documentation and conversational context, returning direct explanations with links to the exact pages that matter.

“We had a customer ask whether someone else could reschedule a meeting. Instead of escalating it, they searched the help center — and AI surfaced the exact answer right away. It felt like having a support desk that fully understands the product running 24/7.”

GitBook Assistant didn’t just answer the question, but it prevented the customer from ever having to contact support. Customers get unblocked instantly, and the team can stay focused on higher-impact work.

Information architecture built for shipping

The immedio team follows one rule: if it’s real, it’s in the docs.

They rebuilt their help center around a simple, predictable structure: release notes, guides and troubleshooting content, and pages mapped 1:1 to their admin screens.

“This structure makes it much easier for customers to find what they need. If you’re in a specific admin screen, there’s a page that matches it exactly.”

That 1:1 mapping removes guesswork during shipping. Every UI screen has a clear home in the documentation, making it easier to keep content current as the product evolves.

Results: fewer escalations and faster iterations

Before the migration, engineering was handling more than 70 escalated support cases per month. After the move to GitBook, inquiries dropped by roughly 30%.

Qualitative feedback matched the numbers. Customers asked fewer questions. Support relied more heavily on AI search. Engineering spent less time re-explaining decisions that were already documented.​

“GitBook helped us turn documentation into something we can actually trust — a single source of truth that scales as the product grows.”

For immedio, the next step is continuing to build on that foundation: keeping docs tightly connected to product changes, and using AI to move faster without sacrificing accuracy.

Want to build documentation that stays accurate, reduces support load, and fits naturally into your team’s workflow?

Start your own GitBook space today and see what AI-powered documentation can look like.