How RTL Accelerated Software Delivery by 8x with GenAI — and What It Means for the Corporate Market

While most large companies are still debating artificial intelligence pilots that never reach production, RTL Netherlands decided to make a different bet — and the results are rewriting what is considered possible in corporate software development. In a report published by Giovanni Piccirilli, former CTO of RTL Netherlands, the world learned the behind-the-scenes story of one of the most successful generative AI transformations ever documented in the media and technology sector.

The result? Deliveries 8 times faster compared to the pre-AI baseline, without additional hiring. But this impressive number is only the surface of a much deeper story about culture, organizational courage, and the true role of GenAI in companies.

---

The Problem No One Wants to Admit: Corporate AI is Failing

Before telling what RTL did right, it is necessary to understand what the market in general is doing wrong. According to Piccirilli himself, most corporate investments in generative AI are not generating concrete business results. The reasons are systemic:

• Companies automate their existing complexity instead of simplifying it
• Pilot projects rarely reach the production environment
• There is significant spending on technology with marginal returns
• The focus is on the tool, not on process change

This diagnosis resonates with what technology sector analysts have been observing globally: organizations that simply plug chatbots or code copilots into dysfunctional processes reap mediocre results. RTL chose another path.

---

The "Moonshot": A Protected Bet with a Defined Deadline

In the second quarter of 2025, RTL's senior leadership (the SLT — Senior Leadership Team) approved a radical experiment in the domain of the Content Supply Chain — the chain that governs how content is produced, cataloged, distributed, and managed within the company.

The format was deliberately radical:

• Duration: 5 weeks in a completely protected sprint
• Stated goal: to achieve a 10x acceleration in feature delivery
• Special condition: the team was freed from all other obligations
• Fundamental cultural premise: failure was explicitly accepted as a possible outcome

This last condition is more important than it seems. In traditional corporate environments, the fear of failure is the primary enemy of real innovation. By openly saying "it might not work and that's okay," RTL's leadership removed the psychological brake that paralyzes most corporate experiments.

---

The Two Pillars of Transformation

The experiment was not based solely on adopting a new AI tool. The transformation was built on two simultaneous structural changes:

1. Redesigning the Operating Model

The previous model was sequential and specialist-dependent: each stage of development passed from hand to hand between different experts, creating bottlenecks, delays, and context loss. RTL replaced this with a collaborative micro-team model, where small, multidisciplinary groups operate with autonomy and end-to-end vision of what they are building.

This change alone would have had a significant impact. But it was combined with the second pillar.

2. Generative AI for Mechanical Coding Work

GenAI was introduced to take over the majority of mechanical coding work — generating repetitive code, documentation, basic tests, and scaffolding structures. With this, humans were freed to focus on:

• Intent — what the software needs to do and why
• Quality — critical review and architectural decisions
• Business decisions — what really matters for the final result

The immediate result of the experiment was a 3.5x improvement in delivery speed — without hiring a single additional person. An impressive number, but one that would prove to be just the beginning.

---

The Most Important Legacy: The Cultural Shift

Piccirilli is emphatic in stating that the greatest legacy of the moonshot was not the 3.5x figure. It was what happened afterward.

More than that, the team left the experience with a new mindset: the willingness to continue changing how they work. In the following months, without the presence of the original CTO, the team went even further:

• Rebuilt the technology and infrastructure from scratch
• Placed AI agents — not humans — at the center of the workflow
• Scaled the model beyond what had been built during the original experiment

The consolidated result: 8x acceleration in delivery compared to the pre-AI baseline.

---

What the Market Can Learn from the RTL Case

The story of RTL Netherlands is not just a success case to put in a PowerPoint presentation. It carries structural lessons for any organization that wants to make GenAI work for real:

✅ Simplify before automating

Automating complex processes only makes the complexity faster. You must redesign the process before applying AI.

✅ Protect the experiment

Teams that have to divide their attention between innovation and day-to-day operations rarely innovate truly. Protected space is a requirement, not a luxury.

✅ Formally accept failure

Explicit permission to fail is not a management weakness — it is the only way to free teams to try things that truly matter.

✅ Focus on cultural change, not the tool

The AI tool is the catalyst. The real transformation is in the team's willingness to question and rebuild their own way of operating, repeatedly.

✅ Put AI agents at the center, not on the periphery

The jump from 3.5x to 8x happened when the team stopped using AI as an assistant and started building an infrastructure where AI agents are protagonists of the workflow.

---

A Signal for the Corporate Sector in 2025 and 2026

Piccirilli's report comes at a crucial moment. After two years of hyperbole surrounding generative AI, the corporate market is beginning to demand concrete results. Boards of large companies are questioning the returns on investments that already total billions globally.

The RTL case offers a clear answer: GenAI generates value when the problem is not the technology, but the courage to change the process and the culture around it.

For technology leaders who are still struggling to get their AI pilots off the ground — or who look at pretty dashboards with no real business impact — Piccirilli's message is direct: stop automating complexity. Start simplifying it.

---

Giovanni Piccirilli was CTO of RTL Netherlands and is the author of the original report published on Medium in July 2026. The article was reviewed by engineering and content operations leaders at RTL NL.