The AI Re-Org

Stop Buying Tools, Start Redesigning and Preparing Your Organization

Oct 28, 2025

The AI Value Gap: By the Numbers

For the past two years, the corporate world has been in a reactive sprint to acquire generative AI. That era is ending, and “pilot purgatory” is setting in.

The reason is simple: teams are now confronting the complex, foundational work they’ve been avoiding. They’re realizing you can’t just buy a tool to transform an organization. You have to map workflows, validate data integrity, and redefine roles. In short, it’s time to stop talking about AI and start rebuilding around it.

The key takeaway:

AI and organizational design are inseparable.

If you haven’t started mapping your workflows, data, and roles, you are already behind.

And this isn’t conjecture. The growing gap between AI ambition and real results is now a documented trend of stalled progress across industries.

The Pilot Purgatory: While adoption is widespread, few companies are achieving scale. One McKinsey survey found that while 45% of finance functions are piloting GenAI, only 6% have successfully scaled those initiatives.
The Skills Chasm: The World Economic Forum’s Future of Jobs Report estimates 39% of workers’ core skills will change by 2030. This demands a concrete redeployment strategy, not just a training budget. Yet, many organizations are asking employees to lead the charge without investing the time or resources to develop the skill sets needed to thrive.
The Managerial Bifurcation: Research from Deloitte and HBR shows AI automating not only routine administrative tasks but increasingly advanced functions, once organizations have the process discipline, data integrity, and systems in place to support it. This shift redefines the manager’s role from a task supervisor to a talent coach, enabling flatter, more agile organizations. As spans of control expand, managers who adapt will see their scope, and likely their compensation, grow. Those who don’t will find their roles obsolete.

The Stakes: Competition is Real

As most organizations stall in pilot mode, tech giants and AI-first companies (Google, Microsoft, Anthropic, xAI, OpenAI, etc.) are actively training their entire workforces on this tech in their day-to-day jobs. They are gaining a clear, compounding competitive advantage. This is no different from how Amazon used its data-centric platform to dominate retail. These companies will begin to take out competitors (including potentially your company) who can’t keep up.

This pressure is now amplified as companies try to capitalize on high valuations and potential rate cuts for exit opportunities. The window to “catch up” is closing. The only way to compete is to start focusing on your operating model right now. You cannot realize the full value of AI-driven efficiency without fundamentally rethinking how work happens.

The Real Starting Point: A “Plumbing-First” Approach

When pursuing efficiency, leaders often start in the wrong place. The three most common missteps are:

Buying tools to chase quick wins, only to realize that adoption and integration depend on upstream process discipline and teams with the capacity to support them.
Building tools internally, only to watch the request disappear into an engineering prioritization roadmap that never sees the light of day.
Jumping straight to redesigning the org chart to squeeze teams and increase efficiencies, without providing the training, infrastructure, or clarity to sustain them.

Each of these reflects a desire to outsource discipline instead of building it. Org design, AI tool buying, or building your own tools are all too big, abstract, and disruptive to be a starting point. They are potential solutions to complex problems, once you’ve first identified the real problem, rather than shortcuts to efficiency.

The real work begins with identifying what’s actually broken, not simply declaring “we need more efficiency.” Hint: Start by looking at your top cost centers and revenue drivers, then ask three questions deeper to uncover the true bottlenecks around them. That’s where meaningful, lasting wins are found.

Once leaders stop chasing quick wins and start fixing the foundation, the real transformation begins.

It reminds me of the old parable: a house built on sand will crumble, but one built on rock will stand against the strongest storms. For an enterprise, AI is the house, but your processes are the foundation. That sturdy, “rock” foundation isn’t the “attractive” new AI tool; it’s the unglamorous, critical work of strong workflow documentation, data integrity, and deep process understanding.

The real starting point, and the only one that builds a scalable, low-risk foundation is disciplined workflow and data mapping. Attempting to redesign an organization before documenting its work is pure guesswork. And automating a workflow you don’t trust is how you scale errors at the speed of light.

This “plumbing-first” approach means documenting the end-to-end process in granular detail:

The Steps: Every action, manual or automated.
The Owners: Who is accountable for each step?
The Systems: Where does the data live, and where does it move?
The Controls: Where are the checks for accuracy and approval?
The Exception Paths: What happens when (not if) it breaks?

This approach isn’t a new exciting AI application, but actually the disciplined application of unexciting (but very important) foundational governance. Frameworks from NIST and COSO simply reiterate what high-performing organizations already know: process mapping, data validation, and robust control design are non-negotiable prerequisites for any scalable deployment, especially with AI.

Before you build AI, fix the plumbing. The ROI starts there.

Share Devon's Substack

A Mini-Case Study: From Silo to “AI Pod”

A mid-sized SaaS firm, with its teams traditionally siloed in “AP” and “GL,” maps its “Procure-to-Pay” process end-to-end for the first time. The discovery is shocking: 29% of all AP effort is spent manually correcting invoice errors, a colossal waste of human capital.

The root cause isn’t one thing, but a systems problem. A polluted and inconsistent vendor master data file, combined with fragmented invoicing and payment systems from multiple acquisitions, means invoices constantly fail to match POs. This has resulted in millions of dollars in delinquent payments and constant internal escalations.

If this company had just bought an AI tool, it would have only automated the chaos.

Instead, they executed the “plumbing-first” approach. They first reviewed all of their workflows and identified the issues that would have the greatest impact if solved. Then they cleaned and locked the vendor data, standardized invoice intake fields, consolidated systems across acquired entities, and defined clear exception-handling rules.

Only then did they deploy an AI model to automate matching and coding, supported by AI-assisted dashboards that leaders could query to understand errors.

To manage this new process, they created a “human-AI pod”, a small, cross-functional team consisting of an AP lead, a Procurement lead, and a Systems lead. This pod owned the entire workflow, not just a single task. Their new job was to run the exception queues daily and audit the AI model for drift weekly. They evolved from “doers” to “orchestrators” of the process.

With clear goals to shorten invoice processing time, eliminate late-payment fees, and remove duplicate payments, the team made immediate, measurable gains. The talent evolution was just as critical: the AP supervisor, freed from chasing invoices, was redeployed to a new, higher-value role focused on vendor risk analytics and cash-flow optimization. This is a perfect example of the upleveling and strategic redeployment required to capture the real value of this trend.

This pattern isn’t unique. Every high-performing transformation follows the same logic: stabilize the system, standardize the data, then scale with AI.

This is the model. They fixed the plumbing, then built the house. Once this “workflow-first” foundation is set, it unlocks the three major structural shifts that define the next-generation operating model.

Once the plumbing is fixed, leaders must re-architect how work, management, and talent evolve.

The Three Structural Shifts of the AI-Ready Organization

Once the “workflow-first” foundation is set, it unlocks the three major structural shifts that define the next-generation operating model. This is how you move from “fixing the plumbing” to “building a new house.”

1. From Functional Silos to End-to-End Workflows

This is the most fundamental and powerful shift. For the last century, we’ve organized work into functional silos, specialized departments that handle one piece of a larger puzzle. This creates friction, miscommunication, and a lack of true ownership.

The AI-ready organization dismantles these silos and reorients teams around end-to-end process ownership. This applies to every core value stream in the business:

Finance: Siloed “AP,” “GL,” and “Revenue” teams become integrated “Procure-to-Pay,” “Record-to-Report,” and “Order-to-Cash” teams.
Go-to-Market: Fragmented “Sales,” “Legal,” and “Deals Desk” teams merge into a unified “Lead-to-Revenue” pod that owns the entire contracting and closing motion.
Strategy: Disconnected “FP&A,” “Sales Ops,” and “Business Unit” teams form a “Forecast-to-Execute” team that owns the entire planning and resource allocation cycle.

These new teams, these “human-AI pods”, are small, autonomous, and cross-functional. Their job is no longer to just execute a task but to own an outcome, manage the AI agents (and review their outputs for hallucination risk and judgmental overrides), handle exceptions, and continuously improve the process.

2. From “Manager” to “Coach” (The Great Bifurcation)

This is the most critical, and misunderstood, structural shift. When AI agents and automated dashboards handle routine monitoring and data aggregation, the manager’s role doesn’t just “evolve”; it bifurcates.

Group 1 (The Eliminated): Managers whose primary value is supervising tasks, chasing status updates, and acting as information conduits between staff and leadership. AI will automate this layer of supervision, and those who rely on it will struggle to maintain their roles.
Group 2 (The Thrivers): Managers who redefine their value around the purely “human-centric” work: coaching talent, developing future leaders, exercising complex judgment, and managing strategic exceptions. These are the true leaders of the future.

In this new reality, transparency replaces politics, and data replaces perception.

This bifurcation introduces a critical new organizational risk: the “political manager” who masters the optics of leadership (Group 2) but in reality continues to operate as a task-supervising conduit (Group 1). Let’s be real: these people already populate much of upper middle management. Now, with remote and hybrid work, they’ve become even harder to spot—and AI will raise that challenge exponentially.

To counter this, senior leaders must use the new tools at their disposal. AI-driven dashboards and unfiltered, quantitative data from the front line will be key to “seeing through the noise” and distinguishing real outcomes from performative leadership.

This leads to the new organizational paradox: companies will need fewer managers, but the ones who remain will be far more valuable. These higher-paid “super-managers” will be the ones capable of leading flatter, more agile organizations, precisely because their focus is on outcomes, not optics.

3. From “Upskilling” to “Strategic Redeployment”

For years, “upskilling” has been a vague and passive buzzword. Most companies provide a library of optional courses, check a box, and then wonder why their workforce isn’t transformed. This “field of dreams” approach fails because it places the full burden of transformation on the individual employee, disconnected from a clear business objective.

The AI-ready model replaces this passive hope with active, strategic redeployment.

This is a proactive, forward-looking workforce planning discipline. It’s not about hoping employees find the right skills; it’s about designing the future roles the business needs and then building the paths to get your people there. It means you identify the new, high-value jobs your organization requires before the automation is even complete.

These target roles are not just new titles; they represent a fundamental shift from “doing” to “orchestrating” and “analyzing”:

Process Owner / Orchestrator: Manages the end-to-end “human-AI pod” workflow, focusing on efficiency, exception handling, and continuous improvement.
AI/Model QA: A critical governance role focused on auditing model outputs for accuracy, bias, and operational drift.
Data Analyst / Strategic Finance Partner: Moves from data aggregation to data interpretation, using AI-driven insights to guide high-stakes business decisions.
Vendor Risk / Exception Manager: A high-judgment role focused on solving the complex, non-standard problems the AI cannot.

As automation predictably frees up capacity from repetitive, low-judgment tasks, you now have a pre-defined, high-value path to move your best people into these new, critical roles.

This is how you capture the true productivity gains of AI. You reallocate your most valuable asset, your people, to the complex, creative, and strategic work that drives the business forward.

Practical Takeaways & How to Engage

STOP starting with technology.
START by mapping your 3-5 most critical end-to-end workflows and auditing their data integrity.
AUDIT YOUR MANAGERS. Identify who is a “task supervisor” (at risk) vs. a “talent coach” (critical for the future).
BUILD a “redeployment map,” not just a “training plan.” Identify the high-judgment roles your team will move into before you automate their current tasks.

Thank you for reading. If you found this helpful, please share it with a colleague or subscribe to the Substack.

My 2026 advisory calendar is nearly full. If your team is looking for an expert to help build this “workflow-first” roadmap, send me a message directly to secure one of the remaining slots.

Thank you for the support and have a great day!

Best,

Devon

Note - I have summarized my insights from the below articles and research. I have not directly cited them to save time in the drafting process, but have included them below in case you are interested in a deeper dive.

Key Research for Reference

For your reference, here is some of the research that informed these insights. The recent McKinsey, HBR, and NIST papers are particularly insightful.

On Culture & Change Management