Agent

Working paper review

First review is where a manager spends an evening writing points that are largely the same points as last time. This agent works through the prepared file against your own review programme and returns the list: figures without support, schedules that disagree, cross-references that go nowhere, treatments that differ from last year without an explanation. The reviewer's judgement is untouched.

How it runs

Step by step

01

Connect the engagement file

The current working papers, the prior year file and the schedules behind them, connected read-only.

02

Load your review programme

The firm's own review points and file standards become the agent's checklist. It brings no methodology of its own.

03

It reads the file through

Every working paper in scope is read, not sampled, and each figure is traced to whatever supports it.

04

It lists the exceptions

Unsupported figures, disagreements between schedules and broken cross-references come back with the paper referenced.

05

It compares with last year

Treatments that changed without an explanation on file are raised as questions, not corrected.

06

The review manager approves

The list goes to the preparer only after the review manager has read it, removed the noise and added their own points.

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

Managers carrying first review across a portfolio, and firms where review points arrive late because the reviewing is done last.

Escalation

When it asks a person

The review manager approves the list before it is raised with the preparer

Applications & Rights

Which tools it uses, and what it may do in each
SharePointRead only
DropboxRead 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

Managers carrying first review across a portfolio, and firms where review points arrive late because the reviewing is done last.

What you’ll need

  • Current and prior year files connected
  • Your review programme supplied
  • A location for the points list
  • A review manager to approve

What you get

  • A list of review points per file
  • A link to the paper behind each point
  • A list of unexplained changes from last year

What it doesn’t do

It is not the reviewer of record. It does not sign a file, form an opinion or take responsibility for the work, and a point it fails to raise does not shift accountability anywhere.

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.

Manual first-pass file reviewReview point lists written by hand
Estimated savingNot setNo figure ships until someone at ollo owns it and the method behind it.

Frequently asked

Does this replace the reviewer?

No, and any tool claiming otherwise should be treated carefully. It does the mechanical pass a reviewer would otherwise do late at night, so the reviewer's time goes on judgement. The file is still signed by a person who has looked at it.

What kind of points does it actually find?

The unglamorous ones: a figure with no supporting schedule, two papers that disagree, a cross-reference pointing at nothing, a treatment that changed silently. These are the points that clog a review and they are exactly the points a careful reader catches.

Can it check arithmetic?

It checks what the documents state against each other. It does not recompute a position or re-perform a calculation in a system it cannot see, and it says so rather than implying a check it did not do.

Where does the list end up?

In a drafting location you choose, and it reaches the preparer only after the manager approves. That ordering matters: a raw machine list handed straight to a junior would be demoralising and often wrong.

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.