The lab · 8 experiments

Working models for the finance function AI makes possible.

Each experiment takes one part of the job — planning, close, consolidation, diligence, treasury — and shows, with the numbers moving, where machine intelligence changes the work and where judgement still has to sign it off. Nothing here is a mock-up. Every figure is computed in your browser.

01Close
The close, compressed
Month-end as a critical path. Switch on automation task by task and watch the close collapse.
Compress the close
02AI automation
Confidence, not magic
An AI categorises a live bank feed. You set the confidence bar — and own what slips under it.
Set the threshold
03FP&A
Pipeline to plan
A driver-based rolling forecast built from pipeline and order book, in three cases at once.
Run the scenarios
04AI automation
The variance writes back
Budget versus actuals in, board-ready commentary out. Tune materiality and audience.
Draft the commentary
05Accounting & tax
One group, many ledgers
Multi-entity, multi-currency consolidation with intercompany eliminations and transfer pricing.
Consolidate the group
06M&A
Quality of earnings
From reported to adjusted EBITDA, with machine-flagged anomalies and an earn-out that has to hold.
Normalise the earnings
07Systems & treasury
Cash, in real time
Wire up ERP, banks, payroll, billing and CRM — and watch the cash forecast sharpen.
Connect the stack
08AI automation
Automate or hire?
The AI business case, costed honestly: implementation, ramp, rework and oversight included.
Test the business case
09 · Open
The next experiment is yours.
A decision you cannot model cleanly, a process you suspect a machine could run. Send it over — the good ones end up here, anonymised.
Propose one

The close, compressed

Confidence, not magic

Pipeline to plan

The variance writes back

One group, many ledgers

Quality of earnings

Cash, in real time

Automate or hire?

The next experiment is yours.