PIPELINEGuide
Checklist: preparing an AI knowledge base
What you should clarify before requesting quotes on AI knowledge base. The more of this is settled beforehand, the more reliable the quotes will be.
Before deciding on an AI knowledge base, a quick assessment is worthwhile: which workflow costs the most time, who will maintain the system later, and whether the necessary content even exists. The complete checklist is further below.
01 — Checklist
Clarify before the first quote
- What should improve? In one sentence, no jargon.
- How will you know? One number that should change.
- Who decides? One person, not a committee.
- What budget is available? A range is enough — it makes quotes comparable instead of arbitrary.
- By when? And what the deadline depends on.
- Which systems are involved? And who has access to them.
- What do you already have? Content, data, preliminary work — this reduces the cost.
02 — Participation
What we need from you
- A contact person who can make decisions — not just forward them.
- Access to the affected systems, in time and not on launch day.
- Real examples from daily business — the messier, the more useful for testing.
- Two to four hours for coordination, distributed over the project duration.
- An answer to the question of what success means. Without it, nothing can be assessed afterward.
03 — Questions
Questions you should ask
- What happens if the project becomes more expensive than estimated — who bears that?
- Who owns data, access, and, if developed, the source code?
- What does operation cost in the second and third year?
- What does the documentation we receive at the end look like?
- How would you measure whether the project was successful?
- Is there a case in which you would advise us against this?
04 — Questions
Frequently asked questions
Our documents are a mess. Will it still work?
To a certain extent, yes — the system handles different formats and structures. What it cannot do is resolve conflicting information. If three versions of a price list are in circulation, you need to clarify which one applies first.
Will everyone see everything?
No. The permissions from your existing systems are carried over. If someone isn't allowed to open a folder today, they won't get answers from it either.
What is RAG?
The abbreviation stands for retrieval-augmented generation. Instead of having the language model answer from memory, your documents are searched first, and only the passages found are used to generate the answer. That's why the answer has a source.
How long does setup take with a large document base?
It depends less on volume than on condition: well-structured repositories with few duplicates are ready in a few weeks, grown repositories with many old versions need a cleanup round first.
Does this have to run in the cloud?
No. For sensitive areas, we use models that run entirely on your own hardware. That costs more to set up and slightly reduces answer quality, but no document leaves the building.
What if we have multiple languages in our document base?
No problem — the search understands questions and documents regardless of language and finds the right answer even when the question and source are written in different languages.
Can we roll this out to more departments step by step?
Yes, that's the usual approach. Start with one department, learn from their questions and results, then connect the next one — rather than releasing everything at once.
05 — Next
Related pages
- AI Knowledge Base — Overview
- AI Knowledge Base — Data Protection
- AI knowledge base — funding
- AI knowledge base — alternatives
- AI knowledge base — common mistakes
- AI Knowledge Base — Requirements
- AI knowledge base — AI disclosure
- AI knowledge base — process
- AI automation
- Document processing with AI
- AI chatbot
- All guides
Next step
Let's talk briefly about whether this makes sense for you.
A twenty-minute conversation is usually enough to clarify whether a project makes sense. If not, we'll say so — that's cheaper for both sides than an offer nobody needs.