Agent

Audit evidence desk

Audit season is mostly retrieval: someone reads a request list and goes hunting through folders and mailboxes. This agent takes the request list, finds the supporting documents across the connected stores, and builds an index showing what exists, where it sits and what is missing. The controller approves the index before anything is shared.

How it runs

Step by step

01

Load the request list

The auditor's list is loaded as the definition of what is needed, item by item, in the wording they used.

02

Search the connected stores

The agent looks across the drives, document stores and mailboxes finance has connected, using the naming your team actually uses.

03

Match documents to requests

Each request gets the documents that answer it, with the location and date shown so a reviewer can confirm the match.

04

Flag missing and ambiguous items

Requests with nothing found, or with several candidate documents, are listed separately for finance to resolve.

05

Assemble the evidence index

The index is drafted in the format your audit file uses, with a line per request and a link per document.

06

The controller approves release

The financial controller reviews the index and decides what is released. The agent shares nothing with the auditor.

The harness

Exactly what this agent can see, touch and change

The same five controls sit behind every ollo agent. These are this one's settings — visible before you build it, not buried in an admin screen afterwards.

Context

What reaches the model

Controllers and finance managers handling an external audit or an internal review, where the same documents get searched for by several people at once.

Escalation

When it asks a person

The financial controller approves the index before anything reaches the auditor

Applications & Rights

Which tools it uses, and what it may do in each
SharePointRead only
DropboxRead only
GmailRead only
Google DriveReadWrites newCreates new records or documents. Never edits, moves or deletes anything that was already there.

Verification

How you know it’s right

Every claim links to the document it came from. A statement the agent cannot cite does not make it into the output — which is what makes the result reviewable in minutes rather than re-read end to end.

Who it’s for

Controllers and finance managers handling an external audit or an internal review, where the same documents get searched for by several people at once.

What you’ll need

  • The auditor request list
  • Connected document stores
  • Agreed retention and access rules
  • A controller to approve release

What you get

  • A cited evidence index
  • A list of missing items
  • A list of requests with ambiguous matches

What it doesn’t do

It finds and indexes documents. It does not decide whether evidence satisfies a request, prepare the audit response, or send anything to an auditor.

How you get it

We build the first one with you

Not a template you configure alone. We sit with your team, build it on real data, and hand over the controls.

01

Scope

One session with the people who actually do the work. We agree what the agent reads, what it may write, and who approves.

02

Co-build

Built on your own data, not a sandbox. You watch it being made, so you know why it behaves the way it does.

03

Handover

You own the controls. Change the context, tighten the rights, move the approval gate — without coming back to us.

What it replaces

Parts of the work currently spread across the categories below. It does not replace any of those products outright.

Folder-by-folder evidence huntsShared spreadsheets tracking requestsDuplicated searching across the team
Estimated savingNot setNo figure ships until someone at ollo owns it and the method behind it.

Frequently asked

Can the auditor use it directly?

That is your decision, and most teams keep it internal. The agent reads what you connect and shares nothing on its own, so anything an auditor sees has passed through the controller first.

What if the same document exists in several versions?

It lists them all with dates and locations and marks the request as ambiguous. Choosing which version to release is a judgement about your own records, so the agent does not make it.

Does this work for internal audit and grant reporting too?

Yes. Anything that arrives as a request list against a document set works the same way. The index format is set during the co-build to match the file your reviewer expects.

How do you stop it reading things it should not?

Rights are set per source when we build it, and the agent inherits the access rules of the person asking. Stores that are out of scope for the audit are simply not connected to this agent.

Let's Build

AI is a capability you build. Let's build it together.

30 minutes with our team and you'll leave with a real plan — not a sales pitch.