Without a First Brain, Is the Second Brain Just Scrum in an AI Package, or SOA on Steroids?

Posted on November 22, 2026

0


What happened to Scrum and agile?

Not long ago, every transformation came with sprints, stand-ups and a certified Scrum Master. Today, many companies are quietly rebelling against agile: some have let go of their agile coaches, gone back to old ways of working, and see little return on 20 years of investment in training and frameworks (University of Maryland, March 2026). Agile’s own practitioners now describe teams that became efficient delivery machines, fast but disconnected from actual business outcomes (Scrum.org).

The problem was never the goal. Agility, an operation that responds to change, was always worth having. The problem is how we came to define it. We took agility to mean responsiveness: how fast each team could deliver its piece. We left out cohesiveness: whether the strands of the operation still move together. Agile’s founders wrote about customer collaboration and responding to change. What most enterprises bought was something else: a framework installed from outside, with procedures based on mapped rather than real-world operations. That framework delivered the first at the expense of the second.

Now the same promise is back with a new name: the AI second brain.

A second brain serves a purpose only when it’s fed by a first brain, an understanding of how the operation actually runs. Without one, it’s nothing more than the Scrum and agile approach in an AI package, or service-oriented architecture (SOA) on steroids. Here’s why:

  • The solution comes first. Agile began once the decision about what to build had been made. A second brain begins with the platform. Neither asks the first question: how does this operation actually run, and does it need this at all?
  • It works from the outside in. Agile brought the same standardized roles and routines into every organization, whatever its way of working. A second brain brings its layers and its agents, and fits the organization to them.
  • It learns the documented operation, not the real one. Agile iterated on what users said about the product. A second brain learns from documents and data. The workarounds, spreadsheets and judgment calls that keep the business running are what neither was built to find.
  • Speed stands in for direction. Agile promised faster delivery; the second brain promises faster decisions. Without knowing where you actually stand, both get you to the wrong place faster.
  • When it fails, the implementation gets the blame. Even the University of Maryland piece concludes that “Agile hasn’t failed,” only the way organizations use it. That’s the defense. But an outside-in method produces outside-in implementations. You can’t separate the two.

We’ve been here before

None of this is new to readers of this archive. The vocabulary changes with each technology wave. The pattern doesn’t.

  • 2007: adapting the practice to the software. In The Change Management Myth, practitioners traced poor compliance to one cause: their company had modified its operations to accommodate the technology. Dangerous Supply Chain Myths, Part 6 asked the question that still applies: are you in the software business or the procurement business? Virginia’s eVA program, in Yes Virginia, Part 1, showed the other path: understand the operation first, and let the technology follow.
  • 2008: the canvas and the camcorder. Bill Gates’s digital nervous system was the second brain of its day. In Does Your Enterprise Have a “Digital Nervous System”?, I wrote that equation-based software tries to confine multiple strands into a single static process, while agent-based modeling first seeks to understand the operating attributes of diverse stakeholders. Days later, in Similarity Heuristics, Iterative Methodologies: an equation-based model paints the operation on a canvas; an agent-based model films it as it moves. A second brain fed only on documents is iterating from a canvas.
  • 2011: process that adapts to people. Answering a prediction that processes would become “more agile” through integrated workflow systems, In the Year 2020 . . . Process set out the linchpin: a process that adapts to the way people actually work, not an illusory ideal to which they must adapt. The Emergence of the Metaprise described the agent-based architecture to match.
  • 2026: the substrate under agile. In Eric Kimberling’s Hidden Cost of Agile in ERP, and the Substrate Underneath It: local delivery speed produces incoherence at the enterprise level. Underneath sits the substrate: the spreadsheet logic, informal approval paths and unresolved decisions every new platform is built on. But the substrate isn’t the enemy. Workarounds exist because the mapped system didn’t fit how the work actually gets done. And many of the maverick practices actually did the job. Each workaround marks a place where the map was wrong. In the AI era, people can create them faster than ever: a shadow AI tool can be set up in an afternoon. Agile builds the new system around the workarounds without asking why they were needed. The first brain asks that question, tracing each workaround back to the gap it was filling. The second brain can’t ask it. It only sees what has been written down, and most workarounds never are.

Every wave left workarounds behind. Illustrative; the 2024 figure is from the Microsoft and LinkedIn 2024 Work Trend Index.

ERP, SOA, e-procurement, agile, and now the AI second brain: each arrived from the outside in, and each was blamed on its implementations when it fell short.

The second brain is SOA on steroids. SOA linked the systems an organization already had; the second brain reasons across them and learns from them. That’s a real step up. But neither reaches what was never in the systems: how the operation actually runs. A more powerful bridge across the same gap doesn’t close it.

The difference is the first brain

The first brain is built inside-out, in Phase 0, by finding out how the operation actually runs before any technology is chosen: who really decides, what is already failing and why, and the knowledge people carry that no system has captured. Phase 0 is the step. The first brain is what it produces. From there, actual outcomes keep testing and extending it.

Here’s what that looks like in practice. In 1998, a services provider on contract to Canada’s Department of National Defence was delivering parts next day 51% of the time, against a contract that required 90%. Everyone concluded procurement needed a faster ordering system. The first question we asked was what time of day the orders came in. Service technicians, measured on call volume, were batching part orders to late afternoon, after supplier and courier cutoffs, with most parts crossing the border from the U.S. No system held that picture. Once it was understood, next-day delivery rose from 51% to 97.3% in three months, before the technology was chosen.

Without it, a second-brain-only AI operating system is a skyscraper without stairs or elevators: impressive from the street, but nothing gets from the ground floor, where the work happens, to the floors where the decisions are made.

This isn’t a case against second brains. Built on a first brain, they’re valuable: the first brain supplies the understanding, and the second brain applies it at scale.

If a second brain pilot has already stalled, start there. Trace it back from the outcome that’s missing. That’s what Pilot Rescue does: HPT@HansenProcurement.com.

This is the lens for the rest of the series. The Hansen First Brain and the McKinsey Second Brain: A Short Q&A and The Hansen First Brain and the Accenture Second Brain: A Q&A are published. Deloitte with Palantir, and Globant, are next.

A second brain is only as good as the first brain that feeds it.

-30-

Posted in: Commentary