Insights and Lessons from SAFe Summit San Diego

New engines, same old factories

One thought that followed us home from SAFe Summit in San Diego is that we are about to walk into another productivity trap. We have new technology but the factories around it look the same. Electric motors took decades to pay off for exactly that reason, and so did the computer. Building software has been slow and expensive, and that has been the constraint. AI is now lifting it. But as Goldratt taught us, constraints do not disappear, they move. The new constraint is us, or more precisely how we still work. That is what has to change.

Our current ways of working evolved around limitations that are now disappearing. Software was slow and expensive to build, and practices were put in place to protect that investment. Now that output is fast and cheap, what was built to protect value starts limiting it instead.

Mik Kersten made this concrete at the Summit, presenting his new book Output to Outcome. When electricity arrived, factories that dropped an electric motor into a steam era layout gained little. The line shafts were still there, and the layout was still dictated by them. The real gains came from new factories designed around what electricity made possible.

Putting an electric motor in a steam factory is how you end up with the AI productivity paradox. McKinsey's State of AI 2026 puts numbers on it. 80 percent say AI has made them more productive individually, but only 37 percent see any positive effect on the bottom line. The gains are real, they just never reach the organization.

Outcome and experiment driven

This is why a new operating model is vital, and why it should be outcome and experiment driven. Clear outcomes make sure we are working on the right things. Experiments tell us how to get there. Neither concept is new, but they matter more than ever for making sure we get the results we want. We can now afford to throw finished software away, which was hard to justify when the investment was large.

What AI-Native SAFe changes

Scaled Agile has realized this and introduced a new operating model in their AI-Native SAFe framework, which brings a lot of changes compared to the core framework.

The two-day PI Planning event has become a one-day PI Outcome Planning event, where teams commit to measurable outcomes instead of breaking down the whole PI into planned output. Let that sink in for a moment. It is a major shift in how an ART plans.

Inspect and Adapt once per PI has been replaced by a two-hour Sense and Respond event every iteration, which also absorbs what used to be the System Demo. Feedback arrives while there is still time to act on it.

Tying it together is the outcome tree, which connects portfolio level strategy down to the outcomes that give teams their day-to-day focus. It carries purpose down and commitment back up.

And there is a new role, the AI Value Architect, which we see as a critical success factor. It bridges business and technology, coaches AI adoption, facilitates AI solution development, and owns the job of turning AI workflows and tools into actual outcomes.

First in the world through the door

To get ahead of this we joined the inaugural "Launching AI-Native ARTs" class with the fantastic Rebecca Davis. Being in the first group in the world to take it was a privilege, and we learned a lot. The course does not just walk through the framework, it goes into how you actually launch an AI-Native ART.

We also came home from San Diego with a Spotlight Award from Scaled Agile, with a citation we are rather proud of: "They have been excellent platform partners with great marketing. Supporting Summits and experts at what they do."

What this means for tooling

What the week made clear is that new ways of working also need new product support. Outcome driven planning and roadmaps, curated data, and ways to see whether the work is paying off. Organizations cannot stay in the game on speed alone. That is what we are building toward at Nooga.

Has your organization made a deliberate change to how you work since adopting AI, or has it mostly been about the technology?

‍