Kenaptic is here
Your content estate already contains the answer your customer is looking for. It is just in the wrong silo, with nothing pointing at it. Today we are launching the layer that fixes that — and asks a human before it changes a single page.
Every company that has been shipping for more than a couple of years ends up with the same shape. Product documentation on one platform. A knowledge base on another. A blog on a third. A community forum, and increasingly a Discord, holding the troubleshooting knowledge that never made it back into any of them. Four teams, four tools, and — this is the part that costs money — almost no links between them.
The reader pays for that. A customer hits a problem, opens the docs, finds the page that describes the feature but not the failure, and stops. The fix exists. It is in a forum thread from eighteen months ago, or a knowledge-base article written by a support engineer who has since moved teams. Nothing in the estate points from where the reader is to where the answer lives.
Why this is worth building a product around
Gartner’s 2024 customer service survey put a number on it that we have not been able to stop thinking about. 73% of customers use self-service at some point in their journey — and they fully resolve only 14% of their service and support issues there. Even for issues the customers themselves described as very simple, only 36% resolved fully (Gartner, August 2024).
The reason given is the part that turned this into a product. In 43% of cases, customers could not find content relevant to their issue. Not that the content was wrong, or missing, or badly written — that they could not find it. That is a navigation failure sitting on top of a content estate which, more often than not, already holds the answer.
What it costs to leave there is well established: Gartner benchmarks the median self-service contact at roughly $1.84 against about $13.50 for an assisted one (Benchmarks to Assess Your Customer Service Costs, February 2024) — close to a sevenfold difference. Every reader who cannot find the page walks out of the cheap channel and into the expensive one.
Closing that gap is not a search-ranking problem. It is a problem of the estate having no connective tissue — and connective tissue is the thing no single-silo tool can build, because each one only sees its own content.
What Kenaptic actually does
It reads every silo you connect — docs, knowledge base, blog, forum, GitHub discussions, repositories with no rendered site at all — and builds one model of how the pages relate. Not “these two documents are similar”, which is nearly useless to a reader, but typed relationships: this page explains that one, this thread troubleshoots that feature, this reference is a prerequisite for that guide. The type is what makes a link worth following, and it is what lets us write a sentence a human would have written.
Then it proposes the links that should exist, with the passages from both pages that justify them, and puts them in front of a person. Approved links are written into the actual page source — a pull request for docs-as-code, the platform’s own API elsewhere — so they are real hyperlinks that readers follow, search engines index and AI assistants can traverse. Not a JavaScript widget that only humans with the right browser ever see.
Two rules we will not be talked out of
A person approves every change. Review is the default everywhere, and it is not a setting we bury. A tool that edits your published pages unattended is a tool you will eventually regret, and the first time it gets something wrong in front of a customer you will turn it off for good. So Kenaptic is the analyst; your team is the editor. (Automatic publishing does exist, per source and above a threshold you set — off until an administrator deliberately turns it on, and carrying a warning we make no attempt to soften.)
Everything is reversible in one click. Each link we add is wrapped in explicit markers, so what we contributed is always distinguishable from what was already there — which means removal is exact, not approximate. Retract a link and the page returns to precisely its previous state. We think the fastest way to earn trust with something that writes into your content is to be genuinely, boringly easy to undo.
There is a second audience now
Documentation is no longer read only by people. Coding agents and answer engines traverse it constantly, and they navigate by following links — an unlinked page is, to them, very nearly an unreachable one. An estate that is well connected at the source is legible to both audiences at once. That is a happy accident of doing the right thing for readers, but it is not a small one.
Start with three silos on the free plan and see what your own estate is hiding. Most first runs surface something the team already suspected and nobody had the evidence for.
Kenaptic proposes; a person decides. Every link is reviewed before it publishes and retractable in one click — see Security.
