NL EN IT
Our work

Real results. Real businesses. Real numbers.

A few examples of our work — just some of what we've shipped over the past few years. We don't do theoretical demos: every project below was built for a real client, solving a real problem, with measurable results. Here's what happens when finance expertise meets engineering talent.

AI automation Exact Online

Smart Booking: from 8 hours to 2 minutes

Invoice processing for SME accounting workflows

Live product: financetool.app →

The problem

A mid-size enterprise was spending an entire working day — every week — on manual invoice processing. Staff would open each invoice PDF, read the values, type them into Exact Online, cross-check with purchase orders, and flag exceptions. 247 invoices per batch. 8 hours of mind-numbing work. Error-prone, soul-crushing, and expensive.

Our approach

We built Smart Booking: an AI-powered system that scans incoming invoices (PDF, email, even photos), extracts all relevant data using large language models with RAG (Retrieval-Augmented Generation), validates against existing records, and books directly to Exact Online via the API. Human approval is required for exceptions and amounts above configurable thresholds.

95%Reduction in processing time
99.2%Accuracy rate
247Invoices per batch
2 minTotal processing time
Python Claude AI (RAG) Exact Online API PostgreSQL Azure WebSockets
Business Intelligence Power BI

95% faster reporting for SME clients

Power BI dashboards for SMEs and accounting firms

The problem

Multiple SME clients and their accounting firms were spending 20+ hours per month on manual reporting: exporting data from Exact Online, copying into Excel, building pivot tables, formatting reports, emailing PDFs. Every month, the same routine. Reports were always late, often contained errors, and were outdated by the time they reached the director's desk.

Our approach

We built custom Power BI data models connected directly to Exact Online, bank feeds, and other data sources via Power Query and SQL. Real-time dashboards replaced static Excel reports. Automated scheduled delivery replaced manual email rounds. Each dashboard was tailored to the specific KPIs the business owner actually cared about — not a generic template.

95%+Reduction in reporting time
~0%Error rate (down from ~5%)
Real-timeData freshness (was: monthly)
20+ hrsSaved per month per client
Power BI Power Query SQL Exact Online Scheduled Refresh
ERP migration Process optimisation

Multi-entity ERP migration: €300K in discrepancies resolved

International trading company (Netherlands & Spain)

The problem

A trading company operating three entities across the Netherlands and Spain was running on Snelstart — a system that couldn't handle multi-entity, multi-currency administration. Years of manual bookkeeping had created €300K in unresolved financial discrepancies. Payment terms averaged weeks. The director had no reliable overview of the company's financial position.

Our approach

Full migration from Snelstart to Exact Online. We started with a forensic analysis of all three administrations, identified and resolved every discrepancy, cleaned the data, and migrated to a properly configured multi-entity Exact Online environment. We also automated the FX position management and implemented strict payment term controls.

€300KDiscrepancies identified and resolved
<10 daysAverage payment term (was: weeks)
€35KFX trading profit generated
3Entities unified in one ERP
Exact Online Snelstart (legacy) Data Migration Scripts FX Management Multi-entity Setup
AI automation WhatsApp integration

WhatsApp expense reporting: snap, send, done

Expense management for field teams

The problem

Field teams hate expense reports. Receipts get lost in wallets and car dashboards. Finance departments chase people for weeks. By the time expenses are booked, the month is already closed. The result: inaccurate records, frustrated employees, and wasted finance hours.

Our approach

We built a WhatsApp integration where team members simply take a photo of their receipt and send it via WhatsApp with a short description. AI reads the receipt (OCR + LLM), extracts the amount, date, vendor, and category, and books it directly to Exact Online. The employee gets instant confirmation. Finance gets clean, categorized entries without lifting a finger.

3 secPer expense (was: 10+ minutes)
0Lost receipts
100%Adoption (it's just WhatsApp)
Real-timeBooking to Exact Online
WhatsApp Business API Claude AI (Vision) Exact Online API Python AWS Lambda
AI sales tool Google Cloud

Lumex Energy Calculator: ROI in 60 seconds

Lead-generation tool for a Dutch renewable energy company

Short walkthrough of the calculator's results page

The problem

Renewable energy is complex. Customers want to know payback time, subsidies, and CO₂ savings before they ever talk to sales. Generic calculators give generic answers — and generic answers don't convert.

Our approach

A production React + TypeScript web app that takes an address and a few inputs and returns a personalised, financially rigorous proposal. It pulls roof and sun data from the Google Solar API, parses Dutch energy labels, models ISDE and Warmtefonds subsidies and the declining salderingsregeling, and uses Gemini for personalised AI advice. Sales gets qualified leads with real numbers; customers know their ROI before the first call.

60 secPersonalised ROI calculated
4Product types modelled
LiveSubsidies, tariffs, feed-in data
AIPersonalised advice via Gemini
React 19 TypeScript Tailwind Node.js / Express Google Solar API Gemini Firebase Google Cloud Run
AI backend RAG & GCP

Always-on AI backbone for Lumex's customer service

Multi-service GCP backend with RAG, sentiment analysis, and CRM sync

The problem

Customer service teams drown in calls and emails. Important signals (an angry customer, an urgent case, a product confusion) get buried. Product knowledge sits scattered across PDFs and people, never quite where you need it.

Our approach

A production GCP system with seven Cloud Functions and two Cloud Run services. Sollit CRM data syncs to BigQuery every 30 minutes. Every incoming email is analysed by Gemini for sentiment, urgency, and category. Every call recording is downloaded, transcribed, and tagged. A RAG-powered chatbot on Vertex AI Discovery Engine answers product questions from the live catalog. Dashboards give the team real-time visibility; executives get the KPIs they actually need.

7+2Cloud Functions + Cloud Run services live
5 minEmail-analysis cadence
21CRM tables synced to BigQuery
100%Calls auto-transcribed & tagged
Python Flask Cloud Functions Cloud Run Vertex AI Search (RAG) Gemini 2.5 Flash BigQuery Snowflake Sollit CRM API Gmail API
iOS Native iPad app

Taxato: appraisals on an iPad, reports out the door

Native iPad app for Dutch professional appraisers

The problem

Appraisers visit clients, document business assets, photograph everything, then spend days back at the office writing the formal report. The friction between the visit and the deliverable kills throughput.

Our approach

A native iPad app in SwiftUI that catalogs assets on-site with photos and protocol-compliant questions, supports five Dutch valuation types (LW/MW/OHW/VW/EW), auto-numbers references (TAX-YYYY-NNNN), and generates the formal PDF report directly from the app. CloudKit syncs across devices so appraisals stay safe and continue across iPad and Mac.

HoursSaved per appraisal
Same-dayFormal PDF deliverable
5Valuation types supported
OfflineFully usable without network
Swift SwiftUI SwiftData CloudKit Custom PDF generator

Your project could be next.

Every case study above started with a single conversation. Tell us what's slowing you down — we'll tell you if we can help.

Start a conversation