One chat for your data and your documents

Here’s a situation most operations teams will recognise. Returns at one warehouse have risen for three weeks running.

The numbers are in the warehouse data. The returns policy, which says what should happen to damaged stock, is a

PDF somewhere on a shared drive. The procedure for raising it with the client is in another folder again.

To understand what’s happening and what to do about it, someone has to open a reporting tool, then a file share, then a search box. Then they have to stitch the answers together in their head.

Two kinds of question, one place to ask

Business questions come in two broad kinds:

• “How much, how many, how often?” These are answered by data: volumes, rates, costs, trends
• “What do we do, what does it say, what’s the rule?” These are answered by documents: policies, procedures, manuals.

Most tools handle one or the other. GYCO handles both in the same chat. You don’t pick a mode or a tool first. You

just ask.

How GYCO decides where to look

For each question, GYCO works out whether it needs your business data or your documents, and answers in the way

that suits it.

• Data questions get a table and a chart, built by running a read-only query on your data.
• Document questions get a written answer drawn from the relevant passages, with numbered citations you can

click to see the source.

Here’s an illustrative conversation:

1. “How many returns did the north warehouse receive each week this quarter?” A weekly table and a line chart.
2. “Which clients account for most of them?” The same data, broken down by client.
3. “What does our returns policy say about damaged goods?” A short answer, with citations to the relevant sections of the policy.
4. “And what’s the procedure for raising it with the client?” The steps, cited to the document they come from.

The conversation carries its context from one question to the next, so you can move between the numbers and the

rules without starting again.

When you need one document, not the whole library

Sometimes the question is about one specific document, such as a manual, a policy or a new procedure. In Data

Retriever, you can open any document you have access to, preview it, and chat with that document alone, so the

answers come only from it.

The same access rules on both sides

Data and documents follow the same principle: you see only what you’re allowed to see. Row-level access limits which business data reaches your answers. Department and clearance levels decide which documents GYCO can draw on for you. Our post on permission-aware AI goes into the detail.

What goes in, and what doesn’t yet

GYCO works from information loaded into it, so here’s what that covers today.

• Documents: PDF, Word and plain-text files. Administrators can upload them directly and set a department and clearance level for each one. During onboarding, our team can also load whole document libraries from your shared drives and cloud storage.
• Business data: spreadsheets (CSV and Excel) and tables from your databases, which our team loads during onboarding.
• Not yet: scanned documents without selectable text, images and slide decks. If an important policy only exists as a scan, convert it first.

Because GYCO works on loaded data rather than reading your source systems live, agree how often the data should

be refreshed when you set it up.

Take the answer with you

Answers don’t have to stay in the chat. You can pin a data answer to My Dashboard, where it re-runs on the latest

data each time the dashboard opens, or download it as a CSV or Excel file. You can export a whole conversation,

answer tables included, to PDF. That makes it easy to share the thinking behind a decision with someone who wasn’t in the chat, whether a colleague taking over a shift or a manager signing off a claim.

Why one chat matters

The case for one chat isn’t just convenience. When the number and the rule sit side by side, the conversation gets to a decision sooner: here’s what happened, here’s what the policy says, here’s who needs to act. Fewer tools also means fewer places for access rules to drift apart, and one place to see where every answer came from.