Finance and admin
- Accounts payable: invoice capture, PO matching, coding, approval routing. Typically 60–80% of AP handling time removed.
- Expense and receipt processing: capture, categorise, policy-check, post.
- Debtor follow-up: reminders, statements, payment-plan triggers with a human on exceptions.
- Month-end pack assembly: pull, reconcile, draft narrative, flag anomalies.
Sales and customer
- Lead enrichment and routing: enrich, score, assign, first-touch draft.
- Quote-to-cash: quote generation, approval, invoicing, follow-up.
- CRM hygiene: dedupe, update, log activities, nudge stale deals.
- Support triage: classify, draft, route; resolve tier-1 from the knowledge base.
Operations and people
- Customer, staff and supplier onboarding: collect, verify, set up across systems.
- Scheduling and dispatch updates: jobs, rosters, customer notifications.
- Compliance evidence: gather, file and chase documents against a register.
- Recruiting admin: screen against the role, schedule, keep candidates informed.
How to pick your first three
Score each on volume per week, hours per item, error cost and how stable the process is. Take the top three that nobody will fight you on. Measure the baseline for two weeks before you build — the number you save is only credible against a number you measured.
A single workflow costs A$6k–A$25k to build and under A$300/month to run. At 20 staff hours a week saved and a A$65 loaded rate, payback is usually inside four months. Our ROI calculator uses your numbers.
Questions people ask
Do we need an agent for any of these?
Not at first. All twelve run as rule-based workflows with model steps for reading and classifying. Agents come later, where judgement across systems is the bottleneck.
What tools do you build on?
Open platforms — n8n, Make, Power Automate — plus Python where needed, in your accounts.
Related: AI workflows · Agent vs automation vs chatbot · AI Readiness Audit