This month the AICPA did something it rarely does with a phrase: It trademarked one. "Human in the Lead" emerged so strongly from the Rise2040 conversations, a process that drew on the input of more than 6,000 accountants, that the institute moved to protect it. Set the legal mechanics aside — the trademark is not the story. The design principle behind it is.
"Human in the lead" is a deliberate correction of "human in the loop." A loop is a checkpoint. A human in the loop reviews what the machine produced and clicks "Approve." A human in the lead sets the purpose, owns the judgment, and is accountable for the outcome. One is a speed bump; the other is a leadership model. The profession just told us, with one carefully chosen phrase, where the value is going to live.
Here is the uncomfortable part. A principle is only as real as the architecture built to support it. You cannot keep humans in the lead with a mission statement and good intentions. You keep them in the lead by redesigning two things most firms have left untouched: your governance and your business model. That is the work of the next decade.
The framework for the agentic era
Over the past year I have written in these pages about applying the Exponential Organization framework to CPA firms. The framework has now advanced. Salim Ismail and his co-authors have moved from ExO 2.0 to ExO 3.0 in a forthcoming book, "The Organizational Singularity." It is the most important strategic lens I have seen for what AI does to a professional services firm, and it reframes the question every managing partner is wrestling with.
Start with Ronald Coase, who in 1937 explained why firms exist at all. Markets are expensive to coordinate, so we internalize coordination inside firms. Accounting firms are, at their core, coordination shells for judgment work. We assemble people, route the work, review it, and stand behind it. AI collapses the cost of coordination toward zero. When monitoring, routing, drafting, and reconciliation can be executed by agents at near-zero cost, the old economic logic of the firm inverts. The firm does not disappear. It persists as an accountability shell: The entity that holds fiduciary responsibility and signs its name to the result. The value migrates from doing the work to assuring the work, from labor leverage to judgment leverage.
That single shift is the reason that governance and the business model can no longer wait.
ExO 3.0 organizes the response into two halves that must move together. "DRIVE" is the intelligence engine: decision architecture, recursive learning, the intelligence stack, value moat, and elastic agency. "SHAPE" is the organizational form: safe autonomy, human architecture, adaptive architecture, purpose control, and ecosystem trust.
The summary Ismail uses is blunt and worth posting on the wall: DRIVE without SHAPE crashes. SHAPE without DRIVE stalls. You need both. Most firms are buying DRIVE, a tool here and an agent there, while leaving SHAPE entirely unaddressed. That is precisely the imbalance that produces the failure rate the manuscript cites, where more than 80% of AI deployments do not deliver. The problem is rarely the technology. It is the absence of the organizational form around it.
Governance keeps humans in the lead
For our profession, "Human in the lead" is not a slogan. It is a governance specification. It maps directly onto the SHAPE side of the framework, and it is where firm leaders should spend their attention first.
Begin with the human-AI decision boundary. Every decision in your firm decomposes into a prediction, which AI does well, and a judgment, which it does not. The discipline is to make that decomposition explicit for each workflow, route the prediction work to agents, keep the judgment with named professionals, and recalibrate the line every quarter as the tools improve. That boundary is the operational definition of human in the lead. It is the audit partner deciding what the anomaly means, not the model deciding for her.
Then there is what Ismail calls the Fiduciary Wedge: the permanent gap between what AI can technically do and what it can be held accountable for. "The algorithm decided" is never an acceptable final answer for an attest opinion, a tax position, or a fiduciary commitment. Our standards already say so. The wedge stays wide in audit and review, narrower in advisory, but it never closes. That gap is not a limitation to engineer away. It is the franchise. It is the reason clients will keep paying for a human name on the work.
Governance also means treating every agent in your firm as a defined role rather than a loose tool. ExO 3.0 specifies eight properties for any agent that touches real work: a charter, a named human owner, an autonomy tier, a permission envelope, a memory boundary, escalation rules, an evaluation suite, and an audit trail.
Notice that this is the language our profession already speaks: charters, owners, permissions, escalation, evidence, and audit trails. We have run internal controls for a century. Applying that same discipline to a workforce of agents is not foreign territory for us. It is our home field, and it is the single greatest advantage we bring to the agentic era. The firms that wrap governance, transparency, and a control plane around their agents will not merely catch up; they will set the standard others are measured against.
The business model must change with it
Governance keeps humans in the lead. The business model determines whether the firm thrives while they do.
The billable hour and the traditional partnership were engineered for a labor-leverage firm, one that made money by putting more hours through more people. When agents absorb the operational work, hours stop being the unit of value, and pricing must follow the judgment, the assurance, and the outcome, rather than the time. This is not a rate adjustment. It is a redefinition of what the client is buying.
ExO 3.0 names the new operating logic "elastic agency:" a single pool of capability that blends human and synthetic, internal and external, and scales up or down without reorganizing the firm. A capability registry and graduated authority begin to replace the rigid org chart.
This is where the agentic workforce reaches well beyond the service lines. It is not only tax and audit delivery that gets redesigned. It is firm leadership and management. Middle management becomes what Ismail calls the exception architect, a role where 80% to 90% of the coordination work moves to agents and what remains is judgment, relationships, mentoring, and resolving ambiguity. The pod leader of 2027 oversees a far wider scope with a fraction of the manual load. Finance, scheduling, marketing, and operations inside your own firm are agentic workforce territory just as much as the client engagement is.
Two cautions belong in every leadership conversation. The first is the value moat. If the intelligence engine driving your advantage is wholly owned by a third party who can raise the price at will, your efficiency gains can be taxed away. Own your data and your orchestration and keep more than one model family in play. The second is what I have called the missing junior loop. Automating entry-level work removes the very rung on which we have always trained future managers and partners. The honest answer is that the profession has not solved this yet. Leaders who design a deliberate apprenticeship for the agentic era, rather than letting it disappear by default, will own a talent advantage that compounds for years.
The execution pattern Ismail recommends is to build at the edge, not in the core. The core has an immune system, the understandable instinct of people to defend the structures they know. Stand up an AI-native way of working alongside the firm, prove it on real engagements, and migrate workflows over as the results speak for themselves. For most small firms, the edge is not a separate unit. It is your next 10 engagements run in a new way.
You are the guide
If this feels like a lot, remember the role you are actually playing. In every strong client relationship, the client is the hero, facing a problem and looking for a way through. You are the guide. You bring empathy and authority, and you hand them a plan. AI does not replace the guide. It equips the guide. It clears the operational underbrush so your judgment, your relationships, and your counsel get more of you, not less.
That is what "Human in the lead" finally means for a firm leader. The machine carries the load. The human carries the relationship and the responsibility.
The AICPA gave the profession a phrase worth protecting. Our job now is to give it an architecture worth building. In future columns I will take DRIVE and SHAPE one characteristic at a time and translate each into the language of a CPA firm, with specific moves you can make. We will start where every client conversation should start, with three questions: What work in your firm is coordination, and what is judgment? Do you have an intelligence stack, or just a drawer full of tools? And are you trying to transform the core, or are you building at the edge?
Answer those honestly and you have the beginning of a plan. The exponential future is already here. Keep your people in the lead, redesign the firm around them, and lead your clients into it.
Think — plan — grow!
