Skip to content
Civware

Your data, ready for AIto answer from

AI answers are only as good as the data behind them. We clean, connect and index your documents and records, so AI finds the current, correct source and can show where its answer came from.

Group
Connect
First step
A data check for one use case
You decide
Which sources count as current
Works with
Drives, wikis, document systems, databases

When it fits

Often the first piece of work, especially for assistants and search.

  1. 01

    The same document exists in several versions, and nobody's sure which is current.

  2. 02

    Knowledge is spread across drives, wikis, tickets and email.

  3. 03

    An AI pilot gave wrong answers because it found old or duplicate content.

What we build

  1. 01A map of where your content and records live, and who owns each source.
  2. 02Clean-up of duplicates and retired versions, agreed with their owners.
  3. 03Links between records that belong together, such as a customer and their contracts.
  4. 04A search index that AI can use, with a source on every result.
  5. 05Refresh and retirement rules, so the index stays current.

How it works

Scattered files go through three steps before AI answers from them.

  1. 01Clean: retire duplicates and old versions.
  2. 02Connect: link records that belong together.
  3. 03Index: make the current version searchable, with its source.
From scattered files to one current sourceFiles from a shared drive, a wiki, a spreadsheet and email go through three steps: old copies are retired, related records are linked, and the content is indexed with sources. The result is one current source: version 3 of the travel policy is current and version 2 is retired.FIG. 01FROM SCATTERED FILES TO ONE SOURCESHARED DRIVEWIKISPREADSHEETEMAIL1CLEANold copies retired2CONNECTrelated records linked3INDEXsearchable, with sourcesONE CURRENT SOURCETravel policy, version 3CURRENTTravel policy, version 2RETIRED
Fig. 01 From scattered files to one current source. Retired versions are taken out of search, so AI answers from the right one.

Safeguards

The risks here are quiet ones: an old document, or a file someone shouldn't see.

Answers from old documents
Retired versions are removed from search, and each source has an owner who confirms it's current.
Indexing what people shouldn't see
Permissions from your document systems carry into the index.
Deleting something still needed
Nothing is removed without the owner's approval, and retired content is archived rather than deleted.
The index drifting out of date
Sources refresh on a schedule, and a failed refresh alerts an owner.

How to start

First engagement · Data check

One use case's data, made ready

It delivers:

  • A map of the sources one use case depends on
  • Clean-up and linking of those sources, agreed with their owners
  • A working search index, with sources on every result
  • Rules for keeping it current
Price
Fixed, agreed before we start
Length
Usually 3 to 6 weeks
Decision pointYou decide which sources count as current, and whether to extend the work to the next use case.

Questions

Do we need a data warehouse first?

No. Most AI use cases need a small, well-kept set of sources, not a full warehouse.

What does indexing mean?

Preparing content so it can be searched by meaning, not just keywords, with every result linking back to its source.

Will you move our data somewhere else?

Usually not. The index can run in your own cloud, next to the data.

Who decides what's current?

The owner of each source. We make the choices visible and easy to make, and they make the call.

Tell us one job you'd like AI to do

We'll tell you honestly whether AI is the right fit, and what a sensible first step looks like.

One job is enough to start.

We use your details to reply to you, as described in our privacy policy.