In conversations with over a dozen executives during the last few months, I've heard the same thing repeated over and over again: planning AI initiatives beyond six months is a waste of time. And these aren't just tech companies. Traditional manufacturing, retail, financial services... all of them.

One client recently scrapped a 2.4 million euro AI roadmap they'd finalized just eight months earlier because the assumptions it was built on had already expired. And that's not poor planning, but a signal that the old planning model itself is broken.

In this article, I'll break down why the traditional multi-year AI strategy no longer works, what's actually happening on the ground in European companies, and the specific framework I use with clients to plan AI initiatives in this environment. After leading AI implementations and conducting strategy audits in EdTech and e-commerce, I've identified patterns that separate organizations making real progress from those stuck in constant "pilot mode."

The Reality Gap

Companies claiming they have a solid multi-year AI strategy are either operating with outdated information or choosing to ignore what's happening around them. And I want to be clear: I'm not advocating for chaos or rushing into AI without direction. Having a plan matters. But what's happened in the last 18 months is unlike anything we've experienced before.

Consider the release cadence: OpenAI shipped GPT-5 and then GPT-5.1 within weeks of each other. Anthropic released an entirely new model family. Google recently pushed Gemini 3.0. Open-source models, while still trailing commercial offerings, are closing the capability gap at a rate nobody predicted two years ago. Products that reshape entire workflows are dropping every few months.

Under these conditions, a rigid multi-year strategy isn't just unhelpful. It's actively harmful. You're anchoring decisions to assumptions that have a shelf life of maybe 90 days. By the time you move from concept to implementation, the technical foundation you planned around may already be obsolete.

Europe's Specific Challenge

When I work with companies in Europe, particularly Germany, I notice a distinct pattern. There's real urgency. Executives see major news outlets delivering content through AI avatars (making them realize that Gen AI is already mainstream). They understand competitors could use AI to do their work cheaper and faster. The pressure is there.

But discussions about AI often get stuck in the concept phase. They start with compliance questions, move to data regulation concerns, and sometimes end with internal resistance before any implementation begins. Meanwhile, companies in the US and China are shipping. They're focused on rapid capability deployment while European organizations debate governance frameworks.

This gap creates a specific challenge for local companies. You're competing against organizations that move faster, and your planning approach needs to account for that reality.

The Speed Paradox

Here's something I find fascinating: AI adoption is simultaneously fast and slow within the same organization.

It's fast when you give individual employees new AI-powered tools. Roll out a coding assistant or a writing tool, and you immediately get power users. People are excited. Adoption happens in days.

But when the same company tries to automate a core business process, everything slows down. Legacy systems need integration. Data pipelines require cleaning. Stakeholders raise concerns. These projects can sometimes take more than a year to reach production.

This paradox is exactly why the six-month cycle matters. You need enough time to execute something meaningful, but not so much time that your assumptions decay before you can validate them.

Why Six Months Specifically

The six-month window isn't arbitrary. It reflects what I call "AI time," where things move three to five times faster than traditional technology cycles.

In a typical enterprise software environment, you might plan 18-24 months ahead with reasonable confidence. The platforms won't fundamentally change. The competitive landscape stays relatively stable. Your assumptions hold.

In AI right now, six months ago feels like ancient history. None of us are even excited about new model releases anymore. That's how normalized rapid change has become. A planning horizon longer than six months means you're committing resources based on a world that won't exist when you try to execute.

What Successful Companies Do Differently

I've spent considerable time studying organizations that are actually making progress with AI. Not the ones generating press releases, but the ones deploying systems that affect their bottom line. A few patterns stand out.

They prioritize value over hype. Every initiative ties back to a specific business outcome with a measurable target. Not "implement an AI assistant" but "reduce customer response time by 40% within Q2 and Q3." This sounds obvious, but you'd be surprised how many AI strategies I've reviewed that lack concrete success metrics.

They invest in their people. Training isn't an afterthought. It's built into the timeline and budget from day one. One client allocated 15% of their AI initiative budget purely to upskilling existing staff. That investment paid back within four months through faster iteration cycles in their development team (they simply were able to deliver more value in the same amount of time as previously).

They start small but move fast. Rather than spending six months on requirements and architecture, they ship something in six to eight weeks and iterate. The first version is usually embarrassingly simple. But it's real, it's in production, and it generates actual learning.

They build solid engineering foundations. Fast iteration only works if your data infrastructure and AI engineering practices can support it. Companies that skip this step end up with impressive demos that never reach production (or break immediately there).

The Three Strategic Bets

Beneath these traits, I've identified three specific bets that form the foundation of an effective AI strategy in this environment. These aren't optional nice-to-haves. They're the minimum viable investments for any organization serious about AI.

Bet 1: AI Coding Assistants. This is reshaping how engineering teams are structured and what individual contributors can accomplish. The productivity implications are significant enough that ignoring this bet puts you at a measurable disadvantage within 12 months.

Bet 2: Observability and Evaluations. When you're iterating quickly on AI systems, you need visibility into what's actually happening. Traditional monitoring approaches don't work for LLM-based applications. Building proper observability infrastructure is your safety net when things move fast.

Bet 3: Talent Transformation. There's a skills gap forming right now, and almost nobody is addressing it systematically. The organizations that figure out how to upskill their existing workforce will have a structural advantage over those trying to hire their way out of the problem.

Each of these bets deserves deeper treatment, which I'll cover in subsequent articles. For now, understand that all three need to be part of your six-month planning cycles.

What You Can Do Already Tomorrow

Pull up your current AI roadmap or strategy document. Find every assumption that depends on specific model capabilities, pricing, or competitive positioning. Mark the date each assumption was last validated. Anything older than 90 days deserves a fresh look. Anything older than six months should be treated as potentially invalid until proven otherwise.

This audit takes maybe two hours, and it'll show you exactly how much of your current plan is built on expired information.

Series Navigation

Next in this series:

• Part 2: Stop Hiring: Get 3x Output From Your Current Dev Team With AI Assistants
• Part 3: The Only Way to Prove Your AI Investment Is Worth It
• Part 4: AI-Native Teams: A Practical Playbook for Engineering Leaders
• Part 5: From Zero to Production-Ready AI in 6 Months

Let's Connect