Solution · Knowledge
The answer already exists. Finding it is the job.
It is in a contract, an old email, a manual, or in one colleague's head. Finding it takes longer than doing the work, and when that colleague is on holiday it does not get found at all. We index what your company already knows into one private place, where staff ask in plain language and get the answer with the source document attached.
Free, 30 minutes, no obligation.
What gets automated
- Indexing drives, wikis, mailboxes and past projects in one private place
- Answering in plain language, in Italian or English
- Attaching the source document and the passage the answer came from
- Respecting the access rights each person already has, question by question
The problem
Knowledge nobody can find is knowledge you do not have.
Every company that has traded for more than a few years is sitting on the answers to most of its own questions. They are just spread across a decade of drives, three generations of naming convention, and the memory of the two people who were there. The cost is not dramatic. It is ten minutes, forty times a week.
Today
- 1Knowledge is spread across drives, inboxes and people
- 2Finding an answer means asking around, which costs two people's time instead of one
- 3When that colleague is away, things stop
- 4New people take months to become useful, and most of that is not the work itself
- 5The same question gets answered from scratch every time somebody new asks it
With Reviamo
- 1Documents, procedures and past projects are indexed in one private place, including the file nobody would have thought to look in
- 2Staff ask in plain language and get an answer with the source document attached, so the answer can be checked rather than believed
- 3Access follows your existing permissions — a person sees only what they could already open, and nothing leaves your control
The measure of a good one is not how much it knows. It is whether it says so when it doesn't.
How it works
Indexed privately, answered with the source, bounded by your permissions.
-
1
Choose
We start by deciding what belongs in it with you. A drive with fifteen years of everything is not a knowledge base; the first pass is agreeing which folders, wikis and mailboxes are actually authoritative, and which are history.
-
2
Index
Documents are read and indexed — contracts, procedures, manuals, offers, past project files, scans included. The index is rebuilt as documents change, so the answer follows the current version of a procedure rather than the copy that was current when we built it.
-
3
Ask
People ask in the way they would ask a colleague, in Italian or in English, from the tools they already have open. There is no query syntax and nothing new to learn, because a knowledge base that requires training is one that gets used for three weeks.
-
4
Cite
Every answer arrives with the document and the passage it came from. If the answer is not in your documents, it says so and stops — it does not compose something plausible out of the nearest three files. An assistant that guesses once is an assistant nobody trusts again.
-
5
Bound
Permissions are enforced at the point of the question, not at the point of the index. A person gets answers only from documents they could already open, which is also what makes it safe to point it at HR and contract folders.
Illustrative run. Every row is a step the automation actually performs; the flagged one is the behaviour that matters most.
Plugs into
Where your company's knowledge already sits
- Documents & drives
- SharePoint, OneDrive, Google Drive, Dropbox, network shares
- Wikis & notes
- Confluence, Notion, an internal site, a folder of Word files
- Mail & chat
- Microsoft 365, Google Workspace, Teams, Slack
- Business records
- Your gestionale, ERP or CRM, by API or by export
Not on the list? If it exports XML, CSV or has an API, it is almost certainly fine. The one exception is SAP — we don't integrate with it, and we say so up front.
Who it's for
-
Companies where one person is the manual
Everyone knows who to ask. That is the risk. It is also the reason nothing gets written down — asking is faster, right up until it isn't.
-
Technical and field teams
Installers and engineers who need the right procedure, tolerance or spare part while standing in front of the machine, not back at the office.
-
Anyone onboarding regularly
If new people take three months to stop interrupting the experienced ones, most of that is search, not learning.
-
Firms with real document obligations
Contracts, certifications and procedures you have to be able to produce on request, and be sure you produced the current version.
Questions
The questions people ask about an internal assistant.
Where do our documents actually go?
They stay where they are. We index them in place — the assistant reads from your SharePoint, your Drive or your server, and the index lives in infrastructure you control, in the EU. We are not moving your contracts into somebody's platform, and you can switch it off without getting your documents back, because they never left.
Does it train on our data?
No. Your documents are used to answer your questions and nothing else. There is a real difference between a model that is trained on your content and one that reads your content at the moment it is asked — this is the second. We put it in writing in the contract rather than asking you to take it on trust.
Can someone get an answer from a document they shouldn't see?
No, and this is the part that decides whether an internal assistant is usable at all. Permissions are checked when the question is asked, against your existing access rules. Somebody who cannot open the salary folder does not get answers from it, does not get told it exists, and does not get a summary of it either.
What happens when it doesn't know?
It says so. If the answer is not in the indexed documents the assistant reports that and stops, rather than assembling something that reads well. This is deliberate and it is the single most important behaviour in the system: an assistant that invents an answer once has to be double-checked forever, at which point it has cost you time instead of saving it.
Does it work in Italian?
Yes, and across languages in both directions — an Italian question finds the answer in an English manual and answers in Italian. That matters more than it sounds like it should, because the documentation from a foreign supplier is usually the part nobody reads.
How long does it take to set up?
We start with a free consultation, then a paid process audit that ends in a fixed price. A first assistant over one well-defined set of documents is usually live within a few weeks; the work is in choosing the sources and checking the answers, not in the technology. If your documents also need to be produced rather than found, that is document generation, and the steps are on how we work.
Related solutions
The documents behind the answers.
Operations
Documents that write themselves
Finding a procedure and producing the paperwork that follows it are two halves of the same problem. This is the half where the document gets written.
Read moreHR
CV screening for a real hiring round
New people take months to become useful mostly because they cannot find anything. Worth fixing before the next hire, not after.
Read moreAll solutions
See everything we automate
Six of the processes we're asked for most, each compared before and after.
Back to services
Free consultation
Tell us which question gets asked most, and who gets asked it.
One recurring question and where its answer lives today. Thirty minutes online is enough to say whether this is worth building for you.