An internal knowledge base that answers, instead of being searched
An internal knowledge base with AI answers staff and partners from your own documents and live data, with the source shown. Two ways to get one: our SaaS platform from €89 a month, or a private build from €20,000.
Manuals, price lists, parts catalogues, procedures, dealer terms: the knowledge a company runs on sits in documents that a wiki does not hold and a shared drive does not find. An internal knowledge base with AI reads those documents and answers the question a colleague, a dealer or a technician actually asked, with the source. This page sets out the two ways to get one, what each costs, and the one question that decides between them.
Last updated: 29 October 2026
An internal knowledge base with AI is a retrieval system over your own documents and data feeds, on a signed-in page, that answers a question in the words it was asked and shows the document it used. It is not a wiki: nobody writes pages for it. It is not a search box: it returns an answer, not ten links. There are two ways to get one. The first is our SaaS platform, SimplyBoost, from €89 a month: upload the documents, sign your users in with a few lines of code, push stock or prices from your systems through the API, done in a week. The second is a private build in your own Azure or AWS tenant or on your own servers, when the documents may not leave your estate or the sources need bespoke connectors: a €2,500 audit, a €20,000 Production-ready MVP on your real corpus, production from €50,000. The question that decides between them is where the documents are allowed to live.
What an internal knowledge base is, and what it is not
Three products share the name. A wiki (Confluence, Notion, Slite, Document360) stores pages people write and other people search; it is as good as its last editor. A document store (SharePoint, the shared drive) holds everything and finds nothing. An assistant that answers leaves the documents where they are, reads the passages that match a question, writes the answer and names the source. The third is what this page is about, and it is the one that works when the knowledge was born as a 400-page manual, a supplier's price list or an ERP export, not as a wiki page.
What it reads: PDFs including scanned ones, Word, text, CSV, intranet pages, and a feed from your own systems. What it must do to be trusted: answer only from those sources, show the source every time, and say when the documents are silent instead of guessing. What it must do to be used: live where people already work, on the intranet, inside the dealer portal, on the WhatsApp line the technicians have open.
Four internal knowledge bases we see in practice
The dealer portal. A European light-vehicle manufacturer with dealers in several countries put the assistant inside its dealer portal. The knowledge: technical manuals per model, the parts catalogue, the price list, warranty terms, the dealer agreement. The stock: pushed from the manufacturer's own system as a knowledge source through the API, up to 200 times a day, so a dealer asking whether a part is available gets today's answer. The sign-in: the portal signs each dealer with a key only the manufacturer holds. A Dutch dealer asks in Dutch, a French one in French, from the same German and English manuals. The manufacturer's developer spent an afternoon on two scripts; the support desk now sees only what the documents could not answer.
The sales desk. Delivery terms, discount rules, the right spec sheet, while the customer is on the phone. The price list is the one in force today, not the one from the last training session.
Field technicians. An installer asks on WhatsApp what error code E47 means on one model and gets the manual's answer and the first check to make. The question never reaches the service desk.
HR and operations. Leave rules, the expense policy, the onboarding list, the safety procedure for one machine. New colleagues ask the assistant; the answer is the policy as written, every time.
Buy the platform or build it privately
Road one: SimplyBoost, our SaaS platform. From €89 a month on the Growth plan, which includes signed-in visitors and the API, with no price per user: 300 dealers cost what 10 colleagues cost. Upload the documents, put the widget on a page behind your login (the identity check is a few lines of PHP or Node, and we hand over the guide), push stock or prices from your system on a schedule, and the assistant is live in a week. Hosted in Europe with a data processing agreement; conversations deleted after 360 days. The right road when the documents may be processed by a European SaaS and the sources are documents plus an export.
Road two: a private build in your own tenant. The same retrieval architecture, deployed in your Azure or AWS subscription or on your own servers, with connectors to the document management system, the ERP and the identity provider, and your own roles and retention rules. A €2,500 audit maps the sources, the access rules and the first use case; a €20,000 Production-ready MVP runs on your real corpus against a test set your own people wrote; production from €50,000 adds roles, logging, monitoring and handover of the code. The right road when the documents may not leave your estate, when the sources need bespoke connectors, or when your security officer needs to see the whole pipeline.
Both roads are built and run by the same team, and both are designed to sit next to the ICT partner or internal IT team you already have: they run the estate, we run the AI part and hand over what you want to own. See the private AI assistant for road two in detail.
How to choose in one meeting
Ask four questions. Where may the documents live? If a European SaaS with a DPA is acceptable, road one; if they must stay in your tenant, road two. What are the sources? PDFs, Word, CSV, intranet pages and a scheduled export: road one. A DMS, an ERP and SharePoint with per-folder permissions: road two. Who asks? Up to a few thousand staff, dealers or partners on a signed-in page or WhatsApp: road one. Everyone in a 5,000-seat organisation with their own permissions mirrored: road two. When? Next week: road one. This quarter, with your change process and an audit: road two.
Most mid-sized companies start on road one, because the documents are the same on both roads and the questions people ask are the same. The Missing answers list on the platform, the questions no document covered, is the best requirements document a road-two project can have.
What an internal knowledge base costs
Road one: €89 a month on Growth with 2,000 AI answers, €169 on Pro with 6,000, a quote above that. No price per user, no setup fee. Your own developer's time: an afternoon for the sign-in script and the stock push.
Road two: a €2,500 audit, a €20,000 Production-ready MVP, production from €50,000, all excluding VAT, all code handed over. Running costs are your cloud bill plus, if you want it, managed AI support for the system we built.
What neither road charges for: writing wiki pages. The knowledge is the documents you already have.
Access, retention and the AI Act
An internal knowledge base holds contracts, prices and sometimes personal data, so four rules apply on both roads. Access: the assistant is reachable only from pages your website or identity provider signs; the public assistant and the internal one are separate agents with separate knowledge. Source: every answer names the document, so a wrong answer is traceable to a wrong document and fixed once. Retention: a stated deletion period (360 days on the platform; yours on a private build) and a way to delete one person's data. Transparency: staff and partners see that they are talking to an AI, which the EU AI Act requires. An internal knowledge base that answers from procedures is not a high-risk system under the Act; one that takes decisions about people would be, and that is not what we build.
Frequently asked questions
What is an internal knowledge base?
How is this different from Confluence or SharePoint?
Can it use live data such as stock or prices?
Should we build it or buy it?
Does this replace our ICT partner?
Is an internal knowledge base allowed under the AI Act?
Have a folder of manuals and a list of questions people keep asking?
Book a conversation. We will tell you in one meeting which road fits, what it costs on that road, and what your own developer or ICT partner needs to do.
Book a conversation