Architecture Drives Change

4 min read

From classical to jazz: The new sound of enterprise systems in the AI era

Why enterprise architecture needs both deterministic control and AI-driven adaptability, and how to know which to use where.

Published on
6 september 2026
Written by
Koen Williame, Wim Paredis

When we talk about enterprise systems in today’s IT landscapes, many of them still sound like classical music.

Every process is composed in advance.
Every step follows the score exactly as written.
And the result is always the same: predictable, repeatable, precise.

That deterministic approach still has its place. It works well in environments where compliance, safety, and stability matter most.

But it starts to break down when the environment changes faster than the system can adapt.

And that is exactly the new rhythm AI is introducing into the enterprise world.

Enter Jazz: AI-native systems

Where traditional enterprise systems sound like classical music, AI-native systems sound more like a jazz ensemble:

They share a common framework (chords, tempo, rhythm), but there’s no fixed script.

They listen, adapt, and improvise off one another in real time.

The outcome is not hardcoded in advance. It emerges from context, interaction and timing.

Just as every jazz performance sounds a little different each night, an AI-native solution adapts its behavior to the moment. It does not break the rules, it applies principles with awareness and flexibility.

Classic vs Jazz: what is the impact on enterprises?

The move from deterministic to adaptive systems goes beyond technology. It marks a deeper change in architectural thinking. Control is no longer the only priority; adaptability and resilience now matter just as much.

For enterprise architects, the real challenge is not choosing between structure and flexibility. The real challenge is knowing when to play classical and when to play jazz.

Best when precision, stability, and repeatability matter most. Ideal for regulated or safety-critical environments where the outcome must be predictable every time.

→ Think core banking, payroll, or manufacturing automation: systems that depend on strict control, traceability, and proven results.

Best when adaptability and learning matter most. In fast-changing environments, these systems respond dynamically to new inputs and shifting context, while operating within clear guardrails.

→ Think demand forecasting, dynamic pricing, or intelligent routing: domains where context shifts too quickly for a fixed script to keep up.

Enterprises increasingly operate in environments that feel more like a jazz club than a concert hall. Markets shift, regulations evolve, and data grows faster than any process can adapt.

“A purely classical approach can no longer keep pace with environments that change faster than the score can be rewritten”

Wim Paredis

Managing Partner @ Archers

The price of improvisation

Improvisation isn’t free. A solo pianist is cheaper to book than a full jazz ensemble, and software works the same way: the more autonomy and coordination you add, the more compute you burn.

The rough economics, for the same 100k interactions:

Basic workflow (deterministic): ~$500/month.

One instrument, playing the score.

Single-agent system: ~$2,000/month.

A soloist who can improvise.

Multi-agent system: ~$7,500/month.

The full ensemble, listening and responding to each other.

(Figures based on several studies, at roughly $0.005 per 1K tokens.)

That’s about 4x the cost for a single agent and 15x for a multi-agent setup. The adaptability is real, but so is the bill, and it scales with how much you let the system improvise.

Which is exactly why you don’t make everything jazz. Pay ensemble prices for a part anyone could play from sheet music, and you’re spending for flexibility and resilience that is never used.

Our view

One approach does not replace the other. One approach does not replace the other. Real strength lies in knowing when to play classical and when to play jazz.

In practice, that comes down to two design choices:

→ Some capabilities should remain hardcoded, ensuring reliability, traceability, and control.
→ Others should be adaptive, behavior-driven, and resilient, continuously learning from context and interaction within clear constraints and governance.

The future of enterprise architecture is not classical or jazz.

It is orchestration, in the architectural sense. Knowing which parts play to the score, knowing which parts improvise, and designing how both perform as one: through clear architecture, semantic clarity and governed integration patterns.

“The future of enterprise architecture lies in systems that can blend both worlds, structured where it counts and adaptive where it matters most.”

Koen Williame

Enterprise IT Architect @ Archers

That is where AI-ready architecture becomes essential. Not by adding intelligence on top of old structures, but by designing systems that can listen, learn and evolve, without losing control. The enterprises that succeed will not be the ones that turn every process into improvisation. They will be the ones that know where structure creates trust, where adaptability creates advantage, and how to make both play in harmony.

This requires deliberate architecture, semantic clarity, and integration patterns that allow intelligence to move across the entire landscape.

We help organizations design that foundation, where structure and adaptability work in harmony to create AI-ready enterprises.