An AI Workflow Design,
in full.
Not consultancy to redo every time, but a reusable playbook. This is the Finance & Administration department example, open and free: what to really automate, with which tools, in which phases and — the most delicate point — with which controls when money is on the line. The exact shape of what we graft into your business.
What to really automate
The pattern of what works: high-volume, rule-governed document and transactional work. AI doesn't sign the accounts and doesn't decide cash strategy — it removes the repetitive admin around the number, where one clerk with automation handles roughly four times the invoices of a manual one.
- Accounts payable: invoice reading and 3-way matching al posto di: Manual data entry + document matching by hand
- Faster accounting close al posto di: Month-end reconciliations and variance analysis
- Cash forecasting and FP&A al posto di: Spreadsheet forecast cycles
- Cash application al posto di: Manual remittance-to-invoice matching
- Anomaly and fraud detection al posto di: Sample-based transaction review
The best-documented forecasting rollouts report an accuracy recovery in the order of 25–35% and early visibility on cash dips, but the available figures are self-reported aggregates from vendors and analysts, not independently verified: read them as direction, not a promise of results. About a third of finance leaders still report little perceived value — almost always because of data quality, not the model.
Two models, not a shopping list
The 2026 finance-AI market splits into two models useful to an SME. The right choice depends on where your data sits and how many hours manual work really eats, not on the most-cited tool. The enterprise suites for close and FP&A (like BlackLine or Anaplan) remain a reference ceiling, not a starting point for an SME.
Vertical accounts-payable automation
Tools that attack the work that really eats the hours — invoice reading, expenses, payments — often with a free entry tier. They let you touch real accounts-payable automation before any quote conversation, and measure the saving on a single process.
Horizontal copilot on the ERP
An assistant that speeds up analysis and reporting inside the tools you already use, with connectors to the ERP. Note: it stays a side-by-side assistant, it doesn't handle high-volume work on its own, and the per-user cost has to be weighed against the hours it actually frees up.
Our default choice for an SME is to try the workflow on vertical accounts-payable automation (where the free entry tier cuts the risk), or on the copilot if you're already inside that ecosystem — and move up to a dedicated tool only when volume or close complexity justifies it. Picking the exact vendor — and the due diligence on its data-processing terms — is part of the graft.
The phases of the graft
Every item in the playbook carries a phase, so halfway through you filter by "what comes next" instead of re-reading the pilots you've already closed.
-
1
Pilot
First 30 daysA single high-volume use case — usually accounts payable (invoice reading) or expense categorisation — a named owner, the usage policy written and the metrics instrumented before you start. This is where the first time savings show up.
-
2
Scale
First 90 daysThe workflow that worked is extended to the accounting close and reconciliations, integrated with the ERP you already use. Granular permissions and the audit trail are consolidated: nothing that touches money ships without a human approval gate.
-
3
Ongoing
Steady stateContinuous monitoring of adoption, data quality and results, with periodic review of vendors and risks. The playbook stays alive: it gets updated at every close, not archived.
The compliance overlay, and who governs it
The human on the money (human-in-the-loop)
It's the most important rule in the whole catalogue, and it counts double in finance: an agent with the autonomy to move money — a payment via API, a ledger entry — doesn't pass an audit without a human-in-the-loop architecture. You need granular permissions and a complete audit trail of every decision, not just a prompt. With us no workflow that touches money ships without a human approval gate built from day one: before ROI, a CFO asks about traceability.
Compliance overlay
The receivables and payables cycle touches personal data of customers and vendors: it raises the same GDPR questions as the rest of the playbook — an impact assessment (DPIA) when needed, data minimisation. Onboarding flows with KYC/AML checks raise the bar further. Automated accounting at SME scale isn't in itself a high-risk category under the EU AI Act, but audit readiness (towards a SOC 2) and the vendor's processing terms require the same due diligence as any software choice.
The AI owner
No big-enterprise Center of Excellence: in an SME one or two people named as AI owner are enough. Five responsibilities stay with them — setting priorities, who approves what (payment gates first of all), enabling the team, reusable standards and monitoring adoption and data quality — without assuming a whole dedicated department.
The same pattern, other departments.
-
Sales
Lead scoring, automated follow-ups, CRM hygiene.
See the full example -
Marketing
Campaign optimisation and copy with brand oversight.
See the full example -
Operations
Process automation, reporting and anomaly detection.
See the full example -
HR & People
Screening, onboarding and an internal knowledge base, with high-risk hiring kept under oversight.
See the full example -
Customer support
Ticket deflection, triage and agent-assist, with answers anchored to sources — no hallucinations.
See the full example
This is the example. We graft yours.
The other departments follow the same pattern. Start from the free assessment to find out where it makes sense to begin, or let's talk directly.
Example for guidance only: it does not constitute legal or tax advice or a compliance assessment.