The information exists. Nobody finds it.
Procedures, manuals, wikis, contracts, old tickets. The more documentation grows, the more time goes into searching. In the end people ask a senior colleague, or go from memory with outdated information.
"I'll ask Mark, he knows."
Seniors spend hours answering the same questions over and over.
"Which version is the right one?"
Three PDFs with the same name, only one up to date.
"The customer is waiting."
Support digs through manuals while the ticket stays open.
An assistant that knows your company
Ask a normal question. Get the answer and the document it comes from.
Answers with sources
Every answer shows the document, page and last update date. Check it in one click.
It says "I don't know"
If the answer isn't in your documents, it doesn't make one up: it says so and points to who to contact.
Everyone sees only what they may see
Sign-in with the company account, permissions by role, team or client.
Quality measured, not promised
Before we start we fix 20–50 real questions with expected answers, and test every change against them.
Where it works best
Internal knowledge
Sample question
“What is the procedure for a complaint over €50,000?”
What changes
Faster onboarding, fewer questions to seniors
Customer support
Sample question
“How do I set up device X with firmware 3.2?”
What changes
Tickets closed sooner, consistent answers
Technical documentation
Sample question
“What is the maximum pressure for model Y?”
What changes
The right value, from the current version
Compliance and legal
Sample question
“What does the policy say about contract archiving?”
What changes
Traceable answer, with the exact reference
Customer chatbot
Sample question
“Do you have a free slot next week?”
What changes
Fewer repetitive calls and emails
It makes the most sense where there are many documents, they change often and people keep asking similar questions. If your documentation is small and stable, a simpler solution is often enough: we'll tell you up front.
Question, answer, sources. All on one screen.
The design follows your brand. The structure stays the same: the person asks, gets the answer and immediately sees the documents behind it.
Chat and search
Natural-language questions or classic search, with history.
Sources panel
Document, page and date for every answer.
Feedback
Users flag wrong answers, you see what's missing from the knowledge base.
Admin panel
Sources, sync status, users, roles, costs and usage.
Quality report
Test-question results over time, so you see every change is an improvement.
From your documents to the answer, in 6 steps
Sources
PDF, Word, Excel, SharePoint, Confluence, Google Drive, website.
Preparation
Cleaning, splitting, removing outdated versions.
Index
Content becomes searchable by meaning, not just by keyword.
Search
For each question it finds the most relevant passages.
Answer
The AI answers only from those passages, with citations.
Interface
Chat on the web, in the intranet, in Teams or Slack.
The AI only receives the few passages relevant to the question, never the whole database. We choose the model (cloud or self-hosted) based on data confidentiality and budget.
Your data stays yours
Data in the EU or on your servers
European hosting, private cloud or on-premise.
Per-user permissions
Everyone gets answers only from documents they can already see.
Logs and audit
Question log, configurable retention.
No training on your data
We use models with enterprise policies.
Encryption
Data encrypted at rest and in transit.
Where you already work
On your website, in the intranet, in Microsoft Teams or Slack, or via API in your systems.
- SharePoint
- OneDrive
- Confluence
- Google Drive
- Word
- Excel
- Jira
- Zendesk
- Website / e-shop
- Databases
Start small. You decide whether to continue.
1. Discovery
1–2 weeksWorkshop, document review, 20–50 real test questions.
What you get
Solution design, cost estimate, success criteria
2. Proof of concept
2–3 weeksOne source connected, search and chat on your data.
What you get
Working demo + first quality measurement
3. Implementation
4–8 weeksMore sources, permissions, full interface, security.
What you get
Live version, documentation, training
4. Operation
ongoingMonitoring, data updates, improvements from feedback.
What you get
Monthly report
Each phase can be ordered on its own. After the proof of concept you decide whether to go on. You get exact timings at the end of Discovery.
Clear pricing, phase by phase
The final price depends on the actual work: you get it after Discovery, once we know the sources, users and permissions.
Complete system from
€5k
- Discovery
- Fixed price
- Proof of concept
- Fixed price
- Implementation
- Fixed price by scope, or hourly
- Operation and support
- Monthly fee (hosting, support, small changes)
- AI and hosting costs
- Pay-as-you-go, passed through at cost, estimated up front
What we need to start
- A contact person who can make decisions, plus 1–2 users for testing.
- Access to sample documents (anonymised if needed).
- A list of typical questions with the correct answers.
- Security and data constraints (GDPR, internal policies, where data must live).
What we don't promise
- We don't write your content. We help you clean and structure it, but the accuracy of your documents stays with you.
- We don't guarantee 100%. We guarantee measured reliability. In critical areas (legal, safety) we always recommend a human check.