AI and automation

The problem

• Staff are already using AI tools, and nobody can say where the data went.
• Procurement, privacy and security teams need clear answers on where data is processed and how AI is used before they can approve anything.
• Pilots have shown promise but never made it into day-to-day work.
• Repetitive, rules-based tasks, such as keying invoices, checking codes and matching remittances, take up skilled people's time.
• The review found the problem, but the fix needs a tool your systems don't have, and the vendor won't build it.
• Funders and contracts now ask suppliers to disclose the AI tools they use.

Four ways in

AI automation

1. Find the right taskWork that is high-volume, rules-based and checkable, where automation saves real time and a person can verify the output.
2. Match the tool to your requirementsNova's in-house models, a leading commercial model, or a platform you already license, agreed with you against your data classification and policies.
3. Keep a person in the loopEvery decision that affects a claim or a payment is reviewed by a person before it takes effect. AI speeds up the work; it doesn't make the final call.
4. Build it into the processThe automation runs inside your existing systems and workflows, as part of daily work rather than a side project.
5. Improve each runResults are checked and fed back, so the work gets faster and more accurate over time.

Where Nova uses it today

• Invoice and document extraction: every line of supplier invoices captured, whatever the layout, and mapped to one standard format. See forensic invoice analysis.
• Claim and coding checks: coding and claims checked against contract rules before they are submitted.
• Remittance matching: what was paid reconciled against what was claimed.
• Analysis and reporting: data cleansed, matched and analysed faster, and first drafts of routine reports prepared for people to review and finalise. See Data and insights.

Custom tools, built and hosted

Nova doesn't stop at a recommendation. When the fix needs a tool, Nova Consulting designs and delivers it, and Nova IT, the Nova group's platform company, builds, hosts and supports it.

1. Start from the processDesigned around the work people actually do and the result it has to produce, often building on a review Nova has already completed.
2. Design the smallest useful toolScope, data flows, connections to your systems, security and users agreed up front.
3. Build and connectBuilt by Nova IT and connected to your existing systems by API or file exchange.
4. Test with your teamYour staff test it on real work before it goes live.
5. Host, support and improveHosted and supported by Nova IT, with backups and access control, and refined as rules, contracts or your needs change.

Tools the Nova group has built include:

• Bill Simple: line-level invoice capture that validates supplier invoices and delivers them to the accounting system as coded draft bills, ready to approve, with AI and human review. See Bill Simple.
• Datasimple: a revenue platform with purpose-built claiming, coding and remittance tooling, in-house AI and human review of every claim. Nova's New Zealand revenue maximisation work runs on it: see Revenue Maximisation on novaconsulting.co.nz.
• Payroll checking tools: checks on leave, night shifts, visa conditions and overpayments, so payroll teams keep checking after a review.
• A people management platform: staff and client records, shift availability and timesheets through a mobile app, and recruitment and onboarding, connected to payroll and invoicing.

AI security and governance

• Hosted where you need it. Nova's in-house models can run onshore in Australia, with the model and where it runs matched to your data classification and policies.
• Your models or ours. Already licensed an AI platform? The work can happen inside your approved tools, wherever they are hosted.
• A human on every decision. A person reviews every decision that affects a claim or a payment.
• Your data stays yours. Client data is never used to train commercial models.
• Every tool on the record. The tools used on your engagement, and where each one runs, are recorded, and Nova can keep an approved AI tooling register for your contract.
• Clear answers for reviewers. Procurement, privacy and security teams get plain answers on where data goes and who checks the output, before anything is approved.

AI training

Practical training for your staff on using AI safely and well: what to use it for, what not to put into it, and how to check what it produces, all set against your own policies. Nova's team creates and delivers training on tools such as Claude, and includes people certified in Anthropic Claude and Microsoft Copilot. Nova trains its own staff the same way.

Work Nova has done

Case studies are anonymised and published with each client's agreement.

Frequently asked questions