Do AI agents need Work Architecture to replace legacy systems

There's an ongoing debate. Will AI-native platforms replace legacy systems? Or simply enhance them?

Some argue that structured enterprise infrastructure, like Salesforce and HubSpot, is becoming obsolete. AI-first systems will dominate. Others believe these structured systems exist for a reason. AI will simply embed into existing workflows.

Both views miss the bigger picture.

The real opportunity isn't in slotting AI into outdated workflows. It's in rethinking work itself. Instead of asking where AI fits in, ask this: What does work look like in an AI-powered world?

Every enterprise is deploying AI. Almost none can see the work they're deploying it into.

The work evolution: from systems to subtasks

AI agents aren't ready to autonomously redesign entire business processes. They handle specific tasks well. But they fall short on complex decision-making. The shift from structured systems to AI-native platforms won't be instant. It takes continuous reengineering of work as AI capabilities grow.

1. Today: AI at the subtask level.
2. The next phase: AI-driven workflows.
3. The future: Fully AI-native work models.

The takeaway: stop getting distracted

The AI vs. legacy systems debate is a distraction. The real challenge isn't which system "wins."

It's whether we're ready to continuously reengineer work as AI evolves.

AI is changing how we think about work. That's the real shift. It requires a mindset change:

• Stop looking for a "final" answer. Work will keep evolving. So must we.
• Reskilling must move in sync with AI. If we don't adapt people alongside technology, they'll be left behind.
• Reinvention is not automation. Reinvention means redesigning work. So humans and AI agents operate together. Now and in the future.

The cost of distraction: work will be redesigned with or without you. If you're not driving the change, you're reacting to it. Too late.

Are you still debating systems? Or are you preparing for the continuous reinvention of work?

From Job Architecture to Work Architecture.

Siobhan 💜