Services · Your own search over company documents

Your own search over company documents, on servers you control.

You pay for a tool that searches your company's files and answers questions from them. We set up an open-source replacement on your own servers, connect it to where your documents live, and keep each person's view limited to what they are allowed to see.

100%
Of released code written by AI, and proven by an eval suite
Permissions
Each person gets answers only from documents they could already open
Checked
Real questions with known answers, run after every change

Answers from your own documents, on your own machines.

Every source connected

Drives, wiki, tickets and chat, with a written connection where none exists.

Permissions carried across

Tested with real accounts before anyone else uses the system.

Documents prepared

Split, labelled and stored so the right passage is found.

A model on your side

The answer can be written on your own server, so document text stays there.

Answers checked

A list of real questions with known correct answers, run after every change.

Kept current

New, changed and deleted documents are picked up on a schedule.

A company search tool reads your documents, your chats and your tickets, and answers questions from them. To do that, it has to read your most private material.

Open projects do the same job on a server you run. Our pages on alternatives to Glean and alternatives to Perplexity list them, among them Onyx, Khoj and AnythingLLM.

What we do

List where your documents live. Shared drives, the wiki, the ticket system, chat. For each source we check whether the project has a ready connection, and we write the connection when it has none.

Carry the permissions across. This is the part that decides whether the system is safe to use. A person must get answers only from documents they could already open. We test this with real accounts before anyone else uses it.

Prepare the documents for search. Files are split, labelled and stored so the right passage can be found. Scanned pages and tables need extra work, and we tell you which of your files those are.

Connect a model. The model writes the answer from the passages the search finds. It can run on your own server, so no document text leaves your machines.

Check the answers. We write a list of real questions with known correct answers from your own documents and run it after every change. A wrong answer written in a confident tone is a main risk in this kind of system, and this list is how it gets caught.

Keep it current. New and changed documents are picked up on a schedule, and deleted documents stop appearing in answers.

What to expect

Search over company documents takes more work to do well than a chat screen. The quality depends on the state of your documents as well as on the software. We show you results on your own files early, before the full setup, so you can judge whether to go on.

Common questions

What is an open-source alternative to Glean?

Our directory lists Onyx, Khoj, AnythingLLM and Open WebUI on its alternatives to Glean page, as projects that search a company's own documents and answer questions from them, with each project's licence. Reveneau chooses one for your sources, sets it up on your servers, and connects it to the places your documents live.

How do you stop people seeing documents they should not see?

The search system copies the permissions from each source, so a person gets answers only from documents they could already open there. Reveneau tests this with real accounts at different levels before the system is opened to staff, and repeats the test after every change. A permission fault is treated as a reason to stop the release.

Which document sources can be connected to company search?

It depends on the project. Each one publishes its own list of ready connections, and Reveneau checks that list against your sources on the day. For a source with no ready connection, we write one, provided the source lets a program read its content. We tell you before the work starts which of your sources need this.

Does document text leave our servers?

Document text stays on your machines when both the search system and the model run there. If you choose a paid model to write the answers, the passages used for each answer are sent to that model's vendor. Reveneau sets up the arrangement you choose and states in writing which text goes where.

How do we know the answers from our documents are correct?

Reveneau writes a list of real questions whose correct answers are known from your own documents, and runs the whole list after every change to the system. You see how many answers were right, which were wrong, and why. Reveneau also sets each answer to show the documents it came from, so a reader can open the source and check.

Can company search read scanned pages and tables?

Scanned pages have to be turned into text first, and tables need separate handling so the rows and columns keep their meaning. Both can be done, and both take extra work. Reveneau samples your files at the start and tells you what share are scans or tables, so the effort is known before the full setup.

How often are new documents picked up?

On a schedule you choose for each source. A ticket system might be read every few minutes and a policy folder once a day. Changed documents replace their old versions, and deleted documents are removed so they stop appearing in answers. Reveneau sets the schedule with you and shows when each source was last read.

How is search over company documents different from a chat assistant?

A chat assistant passes a question to a model and returns the model's answer. Search over company documents first finds the passages in your own files that relate to the question, then has the model write the answer from those passages. The second needs your documents connected, prepared and checked for permissions, which is most of the work.

Other services

Replace a paid AI service with one you own.

You pay an AI vendor by the month, by the seat or by use, and the bill grows with your business. We set up an open-source replacement on your own servers, connect it to your product, and build the software around it. At the end the servers, the data and the code are yours.

Your own AI model server, in place of a bill for every request.

You pay a model vendor for every request your product makes. We set up an open language model on your own machines, give it the request format your code already uses, and test it on your real work before you switch.

Your own company chat assistant, in place of a seat for every person.

You pay a monthly seat for every person who uses a chat assistant. We set up an open-source chat assistant on your own servers, with your company sign-in, and connect it to the models you choose.

Your own automation platform, in place of a bill for every task.

You pay an automation service for every task it runs, and the automations your business depends on live in someone else's account. We set up an open-source automation platform on your own servers and rebuild your automations on it.

Your own voice AI: speech, transcripts and phone agents on your servers.

You pay a voice service for every character it speaks or every minute it listens. We set up open-source speech models on your own hardware, connect them to your product or your phone system, and test them in your language before you switch.

Your own AI coding tools, with your code kept on your machines.

You pay a seat for every engineer who uses an AI coding assistant, and parts of your source code are sent to the vendor with each request. We set up open-source coding tools for your team, connected to a model you choose, including one on your own servers.

What are you building?

We would love to hear about it and see how we can help.

Send us a message
Start a project