Airfy Airfy
Deutsch Talk to us

Airfy Knowledge

Ask your company what it already knows.

Your company’s own AI chat. It reads Confluence, Jira and GitLab and answers every question with evidence. If it finds none, it says so openly instead of guessing.

Request a conversation See how it works

Ask in Slack and in AI tools like Claude Code.

Animated demo: a question in Slack is answered with evidence from Confluence, Jira and GitLab. A second question finds no evidence, and Airfy Knowledge says so openly. A third question comes from Claude Code and is backed by file path and line.
Demo

Example with a fictional company (Nordquell) and fictional data.

  1. 01Today

    The knowledge is there. Just not in one place.

    It’s scattered across the wiki, tickets and code. So everyone searches on their own or asks a colleague. And AI tools that lack this context guess.

  2. 02With Airfy Knowledge

    One place that checks for everyone.

    Airfy Knowledge reads Confluence, Jira and GitLab and answers questions from exactly that material. Every answer comes with its source.

  3. 03For people and AI tools

    Everyone asks the same place.

    Your team asks in Slack; AI tools like Claude Code ask directly. Both get the same answer, with evidence.

The one benefit

Every answer comes with receipts you can check at the source.

No answer without an origin. For every statement, Airfy Knowledge tells you where it’s written, so anyone can check for themselves.

MK
Mara · in Slack

How do we ship a hotfix to production?

Airfy Knowledge

Hotfixes go through their own hotfix/ branch and need a review from the platform team. A manual approval is mandatory before the deploy; the step is defined in the pipeline.

JB
Jonas · in Slack

When are we moving to Postgres 17?

Airfy Knowledge

I can’t find any evidence for that, not in the wiki, the tickets or the code. So I won’t guess.

No evidence foundSearched: Confluence · Jira · GitLabNo guess given

If Airfy Knowledge finds no evidence, it says so openly instead of guessing.

1Confluence · wiki page
Address
wiki.nordquell.example/x/hotfix
Title
Hotfix process
Date
12 Mar 2026

Hotfixes are developed on their own hotfix/* branch.

Every hotfix needs a review by the platform team.

Then merge into main and the release branch.

2Jira · ticket
Address
jira.nordquell.example/browse/OPS-412
Title
OPS-412 Manual approval before prod deploy
Date
4 May 2026

Status: Done

From now on, a manual approval before every prod deploy is mandatory.

3GitLab · code
File path
platform/deploy/.gitlab-ci.yml
Line
88
85deploy_production:86  stage: deploy87  environment: production88  when: manual89  only:90    - main

AI amplifies your team. It replaces no one.

The knowledge comes from your team. Airfy Knowledge makes it findable for every person and every tool, with evidence.

Where people ask

Where your team already works.

TodayOutlookOutlook means: on the roadmap, not yet available.

  • Today

    Slack

    Ask where your team already talks. The answer comes with evidence.

  • Today

    Claude Code

    AI tools like Claude Code get evidence-backed context instead of guessing.

  • Outlook

    Web chat

    Ask in the browser.

  • Outlook

    Desktop app

    As its own app on your computer.

  • Outlook

    Airfy app

    On the iPhone.

For IT and data protection

Which data goes where.

IT and data protection have a say. So we show the path of your data openly, step by step.

With answer drafting by a model provider

Airfy Knowledge reads your sources on your server and finds the evidence there. Only to phrase the answer, your question and the evidence excerpts go to a model provider, through your company’s own account under a data processing agreement. No outside AI service reads the documents themselves.

  • Read-only

    Airfy reads Confluence, Jira and GitLab but never writes back. Your IT approves every connection beforehand.

  • Who can ask

    Today, a named group of readers can ask.

    OutlookPer-person permissions: everyone sees only what they may also see in Confluence, Jira and GitLab.

  • Answer drafting

    Today a model provider phrases the answer from the question and evidence excerpts, through your company’s account under a data processing agreement. Without that step: a reading version of the passages.

  • No outside AI in your documents

    No outside AI service reads your documents. Reading and searching happen on the server under your control.

Where the server runs

Always under your company’s control. You choose:

  1. APurchasedThe server is included in the package.
  2. BRented in EuropeA server for your company only.
  3. CYour own serverAirfy Knowledge runs on your premises.

If you prefer, Airfy runs the whole thing as a managed service.

Managed AI

Your own agents for your workflows.

When you need more than questions and answers: Airfy builds custom agents for your workflows together with your team and looks after them, each as a separate engagement.

  1. Built together

    With your team, for your workflows.

  2. Looked after by Airfy

    Also once the agent is running.

  3. Each its own engagement

    Clearly scoped, commissioned one by one.

Getting started

Start small. Decide clearly.

01

Workshop day

A day together with your team as a starting point: your knowledge, your questions, your workflows.

02

Pilot with a clear exit rule

Try Airfy Knowledge in day-to-day work. When and how you can exit is agreed from the start.

We quote prices in the offer.

Status today: Airfy Knowledge is a time-limited demonstration, not in production use.

Outlook

The whole system on a server under your control. Without an outside AI provider.

That’s where this is heading. These items are on the roadmap, so we list them here explicitly as an outlook:

  • Answer drafting entirely on your own server
  • GPU for rent from Airfy
  • Per-person permissions
  • Web chat in the browser
  • Desktop app
  • Airfy app on iPhone

Let’s talk about your company knowledge.

Write to us. We’ll show you the demo and discuss whether a workshop day or a pilot suits you.

Request a conversation

or write directly to steffen@airfy.com