Document capture
Supplier invoices, delivery dockets and statements read with Azure AI Document Intelligence and coded straight into your systems.
Pedersen Group is a data, software and AI consultancy in Canterbury. We help New Zealand organisations, from owner-operated businesses to listed companies, turn operational and financial data into insight they can act on, processes that run themselves, and AI that works on real business data.
in cash found tied up in excess stock and overdue invoices at a pumps and tanks distributor
Read the case studyfaster weekly payroll reporting for a group of companies, with exactly the same numbers
Read the case studyfarm records, from milk production to financials, on an advisory platform we moved 101 legal entities onto
Read the case studyTrusted by organisations across New Zealand














Most AI projects stall because the data underneath isn't ready. We start there, with clean models, secure access and clear ownership, then put AI to work where it saves real hours, with every answer traceable to its source. Already invested in AI and not seeing the return? We fix that too.
Explore AI and agentsAsk your data
Illustration with sample data
Which customers haven't ordered in 90 days but still owe us more than $10,000?
Three customers match. Together they owe $48,200, and the oldest balance is 112 days overdue.
| Customer | Owing | Last order |
|---|---|---|
| Rural supplies, Ashburton | $21,400 | 131 days |
| Irrigation contractor, Timaru | $15,300 | 104 days |
| Civil works, Rangiora | $11,500 | 97 days |
Sources: accounting receivables and sales orders
Supplier invoices, delivery dockets and statements read with Azure AI Document Intelligence and coded straight into your systems.
Plain-English questions answered from your own reporting model, with every answer traceable to the numbers behind it.
Agents that triage the inbox, chase overdue debtors or flag reconciliation exceptions, with a person approving anything that matters.
Your playbooks, review standards and processes written as reusable AI skills, so every agent works the way your team does.
Our own delivery runs on a multi-agent coding workflow. Each agent has one job and a written standard to follow. People make the calls that matter.
A client emails a change. An agent reviews it and flags us if anything is unclear.
An agent writes the specification and an impact assessment.
A developer agent makes the change and writes its tests.
A reviewer agent checks the work against our written standards.
A documentation agent updates the help guide.
A person approves every release. Anything touching master data or database tables comes straight to us, and no agent can deploy.
Veil
Built in-house, now in developmentVeil finds names, account numbers and other identifying details in client files, swaps them for placeholder labels and sends only the cleaned text to the AI model. Authorised staff can restore the real values afterwards. Built for accountants, lawyers and advisers.
What your team uploads
Engagement letter for Jane Smith, IRD 123-456-789, re: Smith Family Trust
What the AI model sees
Engagement letter for CLIENT_0042, IRD TAXID_0007, re: TRUST_0003
A working session where we point AI at your own reports and documents, so you see what's real before you spend anything.
Copilot licences nobody uses, a chatbot that makes things up, a pilot that never reached production. We find out why and get it paying its way.
Data quality, permissions and Privacy Act obligations checked before AI goes live, plus prompt injection and data leakage testing on the AI you already run.
The AI coding workflow we use ourselves, set up for your development team: agents, skills, review gates and guardrails.
Most clients start with one problem: a report nobody trusts, a process that eats a day a week, or a system that won't talk to the others. We fix that first, then help with the next thing.
Your operational and financial data joined up and turned into insight you can act on: live dashboards, board packs and forecasts from one trusted source.
A modern cloud data platform that brings every system into one governed model, ready for reporting, forecasting and AI.
Accounting, inventory, payroll and job systems working as one, with event-driven workflows taking the re-keying, chasing and checking off your team.
Portals, platforms and SaaS products built end to end on .NET, React and Azure, with AI-assisted delivery that puts working software in your hands early.
Agents, copilots and custom skills that work on your real data, plus audits and fixes for AI rollouts that haven't paid off.
Senior technology leadership without the full-time hire: strategy, process and systems design, security, and keeping your software spend honest.
Much of the cost in a growing business sits between its systems: the exports, the re-keying, the spreadsheet that reconciles one to the other. We build the layer that joins them, so data moves once and every team works from the same numbers.
Once it's joined up, your data can answer the questions that matter: where cash is tied up, which customers and products make money, and which assets are sitting idle.
Six sectors where we've done the work, learned the systems and met the people who run them.
Production, payroll and plant reporting across sites and shifts.
Stock, cashflow and freight joined up across every system.
Farm reporting, financial models and advisory platforms.
Jobs, estimating, health and safety, and timesheets in one place.
Yard and plant systems, process maps, platform selection and bin tracking.
Client data platforms, entity structures and group reporting.
Pumps and tanks distributor
We found over $1.5M tied up in excess stock and overdue invoices, then built a cashflow forecast that schedules every bill against the overdraft limit and writes planned payment dates straight back to Xero.
Group of companies
We rebuilt the SQL behind an automated weekly payroll report for a multi-company group. It now runs about 155 times faster, with exactly the same numbers.
Metal recycler
A side-by-side assessment of two specialist recycling platforms, BPMN process maps and a migration plan, plus a bin register that found six lost bins and lifted bin turnover by 12%.
Fixed-scope projects, monthly retainers or fractional CTO advisory. Most engagements follow the same three steps.
A short, paid discovery. We map your systems and processes, and check what you pay for against what your team actually uses.
We start with the change that saves the most time or removes the most risk, so the work pays for itself early.
Then we can stay on as your technology team on a monthly retainer, shipping improvements every month.
Delivered into your own Microsoft and Azure tenant, under your accounts, with documentation another developer could pick up.
Role-based access, secrets in Key Vault and managed identities, with every build reviewed against a written standard.
We’re Kiwi, and so are our clients. GST, KiwiSaver, ACC and daylight saving are second nature, so you won’t be teaching an overseas vendor the basics.
AI
New models from Anthropic and Google work longer and cost less, while older models are retired. What that means for businesses running AI agents.
AI
The newest AI models can work for hours unsupervised. Why most business agents should run inside agent flows, and where letting the agent decide is fine.
AI
Copilot licences nobody uses and answers nobody trusts? The five usual causes of a stalled Microsoft 365 Copilot rollout, and a practical plan to fix it.
Products we build and run ourselves, shaped by years of client work.
AI document privacy for accountants, lawyers and advisers, so client details never reach the AI model.
See how Veil worksAn operations layer for wholesale distributors, sitting over the accounting, inventory, payroll and freight systems they already use.
Ask about early accessConstruction management: CRM, client portal, projects, estimating, health and safety, and timesheets.
Ask about BuildFlowOur energy venture: zero-upfront solar for commercial buildings.
About WattlyTell us about the report nobody trusts or the process that eats a day a week. We'll tell you straight whether we can help, and what it would take.