// four weeks · one process · your data

A four-week proof of concept on one of your own processes

Not a sandbox demo. We take one finance process, run its arithmetic as deterministic code, route explanations to the accountable owners, and measure the result against the baseline you record in week one.

Request a proof of conceptread_only_access · fixed_scope · measured_result
01

What's in scope

one_processOne finance process, one entity set, read-only data access
deterministic_calcDeterministic calculation engine for every figure in scope
governed_loopOwner loop with approval gates and a full audit trail
measurementBaseline versus result measurement, documented
evidence_packEvidence pack you can hand to internal audit
02

The four weeks

The sequence is fixed so the result is comparable. Each week ends with something you can review rather than a status update.

poc · run sequence · 4 steps
00
week_1

Scope and evidence baseline

We agree one process — budget versus actual, a variance investigation, or a rolling forecast cycle — and record how it runs today: hours, hand-offs, and where the numbers currently come from.

✓ baseline recorded
01
week_2

Connect the sources, run the calculations as code

Read-only access to the general ledger extract, budget, and FX rates. Arithmetic runs as deterministic code, not as model output, so every figure can be traced back to a source row.

✓ figures traceable
02
week_3

Governed loop with named owners

Explanations are requested from the accountable cost-centre owners, recorded against the assumption they change, and approval-gated. Disagreements stay visible instead of being averaged away.

✓ approvals logged
03
week_4

Measured result and decision

Side-by-side comparison against your baseline: cycle time, analyst hours, traceability of every published figure, and unresolved assumptions. You keep the evidence pack either way.

✓ result measured
03

What you get at the end

evidence_pack

An evidence pack, not a slide deck

Every published figure with its source rows, the calculation that produced it, the owner who explained it, and the approval that released it.

baseline_vs_result

Baseline versus result

The week-one baseline you recorded yourself, set against what the governed run produced: cycle time, analyst hours, rework.

open_assumptions

Open assumptions, stated

What remained unresolved and why. We publish the gaps rather than smoothing them, because that is what internal audit will ask about.

decision

A decision either way

Continue, extend to a second process, or stop. The pack is yours regardless of the outcome, with no production commitment attached.

04

Request a proof of concept

Tell us the process and the systems it touches. The request goes straight to our team at sales@myoctopus.ai — no mail app, no attachments, no sales sequence.

reply_time — within one business day
first_call — week-one scope and baseline
access_needed — read-only extracts only

We use your details only to respond about a proof of concept. By submitting you acknowledge the scope above: one process, read-only access, arithmetic run as deterministic code, and a measured comparison against your current baseline. See our Privacy Policy.