Technologie / Alphatax
Tech Stack Spotlight
Meer snelheid vraagt meer controle.Voor het Pillar Two-platform combineert Alphatax deze technologieën.
Bekijk de tech stack Van platform tot AI: zes onderdelen van de technische stack.
- Platform
- Microsoft Azure, met .NET en Vue 3
- Infrastructuur en uitrol
- Terraform en Azure DevOps voor automatisering
- Toegang en API-beheer
- Entra ID, Key Vault, private endpoints en API Management
- Monitoring en log analytics
- Application Insights en Log Analytics
- AI bij ontwikkeling
- Anthropic Claude Code voor dagelijks ontwikkelwerk
- AI bij support
- Claude-agent voor ticketanalyse, logonderzoek en het voorbereiden van pull requests
Impact
Service at turbo speed
Scalable product
Four development teams work in parallel on about 30 releases per day, without compromising on quality.
Reliable
Front Leader
Of all software providers, Alphatax has served the most clients with their Pillar Two filings.
About
Alphatax develops software for the tax departments of multinational companies. With Alphatax OS, the company is building a single Tax Control Framework to address complex tax issues worldwide. The company operates in the field of multinational taxation, serving everyone from the small multinational around the corner to the largest in the world.
Behind that ambition lies pressure as well. Alphatax experienced strong growth. It was precisely during that growth phase that the organization faced its toughest test yet: a completely new global tax law, with a deadline that no one could postpone.
“In an AI-first environment, my focus shifts to the desired outcome and the overall quality. You keep that in mind even during the preparation phase.”
THE LEGISLATION WAS IN PLACE. THE TAX AUTHORITIES THEMSELVES WERE NOT YET READY.
Sander Smolders, VP of Engineering International Tax at Alphatax, calls Pillar Two the biggest transition in the tax system since the introduction of the euro. There is now a global minimum effective tax rate of 15 percent, supported by the OECD and G20 and adopted by more than 140 countries. To demonstrate compliance with this threshold, every entity within a multinational must submit 385 data points. For multinationals, having 1,500 to 3,000 entities is not uncommon, and a tax return can quickly run to thousands of pages—in one client’s case, even more than 4,000.
The deadline for the first filing was set for June 30, 2026, with no room to reduce the scope. Alphatax not only had to submit the filings on time; the company was also in the process of developing a Pillar Two product. The biggest challenge came from the implementing authorities: tax agencies in countries that weren’t ready on time themselves. Right up until the last week, they were changing their specifications and interpreting the same legislation slightly differently.
Yannick Peelen, Rockstar and Senior Software Engineer: “As soon as we have the input, we can hit the gas. It’s just that uncertainty from external factors—that’s the tricky part.” The only thing that was certain was that the law wouldn’t wait.
THE SOONER THE VERDICT IS REACHED, THE HARDER THE PUNISHMENT
Team Rockstars provided engineers who could bridge the gaps between technology, product, customers, and business. “Seniority was a deliberate choice,” says Yannick. “If we have to hold someone’s hand, we don’t have time for that.” According to Sander Smolders of Alphatax, software engineers need to think more from a product perspective, and product managers need to understand software better. In an AI environment, those roles are becoming increasingly intertwined. That requires ownership, autonomy, and the ability to organize discussions with tax experts on your own.
According to Sander, Yannick was assigned one of the toughest edge cases right before the deadline. “But I was completely confident that he would solve it without any trouble.” A day later, the fix went live, after completing the full cycle of QA, validation, and approval.
THE HARNESS THAT DRIVES SPEED
Alphatax developed its own AI harness. Domain knowledge, conventions, and quality requirements are embedded in the codebase, with mutation testing and red-green gating that define in advance what constitutes a successful outcome. As a result, AI agents can independently build new countries, form sections, and integrations, with automated tests and human validation serving as the gateway to production.
Within those boundaries, AI performs work in parallel. The turnaround time for new functionality and integrations dropped from weeks or months to days. Four teams collectively delivered approximately thirty releases per day.
LIVE LEGISLATION PROGRAMMING
A last-minute update from the British tax authority goes live within an hour. “That’s live legislation programming,” says Sander. “It’s only possible with AI and the right people.”
Two weeks before the deadline, the number of service tickets rose from 5 to 100 per day, with only two service desk staff members. Sander deployed an internal AI agent that sorted tickets, examined logs and code, and—in the case of a well-defined bug—could prepare a solution and a pull request on its own. An engineer remains responsible for approval. “That has a direct impact on our customer satisfaction,” says Sander.
MARKET LEAD
Sander compares the final phase to a Formula 1 sprint, with the corresponding quality standards. This did not go unnoticed in the market: “In the last two weeks before the deadline, competitors lost customers to us.” Even the onboarding process went at top speed: one to two weeks, compared to the usual three to five months. Other top results:
- Market Leader: Of all software providers, Alphatax has served the most customers with their Pillar Two filings.
- Lightning-fast customer service: bug fixes reduced from days to 30 minutes—from report to production—thanks to AI triage at the service desk.
- Scalable product: Four development teams work in parallel on about 30 releases per day, without compromising on quality.
- Reliable: servedmore than 500 multinationals on time and submitted thousands of returns, despite legislative interpretations that changed right up until the last week.
Collaboration
“If I hadn’t had these Rockstars on board, this project wouldn’t have succeeded.” Alphatax provided the AI vision, tax experts, product knowledge, and the harness. Team Rockstars IT supplied senior engineers who worked independently and helped develop the methodology. Yannick: “My focus shifts to the desired outcome and the overall quality—even during the preparation phase.”
That alignment isn’t a given. Sander selected team members based on ownership and autonomy. The strength of Team Rockstars IT: a culture and talent pool with the profile that an AI-first organization needs. “At their core, they’re mini-entrepreneurs.” AI is constantly opening up new avenues. That requires engineers who can oversee the customer, the business, the product, and the technology—and make lightning-fast decisions without having to go back to a manager.
Sander calls that culture “gold”: people who take on responsibility and become part of the team. That offers Alphatax more than just development capacity: “People in strategic roles are given the freedom to think strategically. Because the day-to-day operations are in good hands.”
