Lead at Scale · White Paper · September 2026Management by AI
The New Best Practice Leaders Need to Master
Management by AI (MBAI) is not the abdication it sounds like. It is handing off discrete parts of the management function — first-pass review, coaching on the gap, keeping the operating rhythm honest — to AI that carries your standards, so your attention goes where only a manager's can.
The TL;DR (short version)
Every generation of management gets a shorthand for where the leader's attention should go. Management by Objectives put it on outcomes. Management by Walking Around put it on presence. Nobody took either literally. MBWA never meant you could run a company by chatting at people's desks.
MBAI is the next one. AI is now good enough to carry select, bounded pieces of the management job, and a leader who hasn't handed those pieces over is spending scarce attention on work a tool does well. Most leaders suspect this already.
The test is narrow. Write the standard down, give the AI a discrete task with a clear input and a clear output, keep the decision with a human, and measure what comes back.
A term with a lineage
MBO, MBWA, and now MBAI.
Each shorthand names where the leader’s attention should go.
Management shorthand tends to name whatever leaders were neglecting at the time. Drucker's Management by Objectives named outcomes, at a moment when activity was being mistaken for progress. Management by Walking Around named proximity, after a generation of managers had retreated into reporting lines.
Both were correctives. Neither was a complete method, and nobody sensible treated them as one. MBWA never asked managers to abandon strategy for permanent circulation; it observed that part of the job only happens in person. The point was the reallocation of attention.
MBAI is the same shape of correction. The neglected question is which parts of the management job to stop doing by hand.
(One housekeeping note: Google MBAI and you will get MBA admissions tools and "MBA with AI" degrees from business schools in India. This is neither.)
The use cases are manifold
MBAI is bigger than document review. It spans the work a good chief of staff does, the review a good editor does, and the coaching a good mentor does. The current portfolio:
- Deliverable review and coaching: a first pass against a written standard, scored dimension by dimension, with specific feedback on the gap — as many times as it takes, at any hour, without losing patience.
- OKR and execution automation: goals updated from the work already happening, streams kept aligned, disconnects and off-track activity flagged early. Team members can ask how their work fits with other streams instead of finding out at the quarterly review.
- Strategy discipline: at the leadership and board level, a check that decisions trace back to the strategy. For individual managers, a resource they can query to test whether they actually get it, without spending the CEO's time to find out.
- Notetaking and follow-through: meetings that write their own notes, decisions that get recorded, actions that get chased. The most widely adopted MBAI practice today, and many leaders don't yet call it management.
- Reading the room at scale: querying meeting notes and open channels to gauge how leadership ideas are landing — where there's genuine buy-in, where the same objection keeps surfacing, and where there's polite silence.
The last one carries a warning label, and the label has a name. Erudit raised over $10M to infer burnout, engagement, and friction from employee communications, and was out of business by April 2025 after the EU AI Act prohibited inferring emotions in the workplace and US states tightened rules on AI-driven employee monitoring. The line it crossed is the one that matters: analyze what people say about the work, in open work channels, in aggregate. Never private communications, never a dossier on a person, and never their inferred emotional state. On one side of that line is listening; on the other is surveillance, and regulators have now picked a side too.
- What people say about the work
- Open work channels
- In aggregate
- Private communications
- A dossier on a person
- Inferred emotional state
Why MBAI arrives as software, not a chatbot
A general-purpose chatbot can do pieces of this on a good day, and leaders already improvise it with CustomGPTs and Claude Projects. What turns improvisation into a management practice is structure: the standard encoded once, distributed to a team, applied the same way every time, and measured. The discipline and customization of AI, with an overlay of SaaS structure.
Full disclosure: Clayton, the signature below, is one of these products. A leader builds a Clayton, a review agent that carries their standards and chosen approaches, and shares it with the people whose work they review. Team members submit drafts and get back a scored, rubric-based assessment with specific coaching, before anything reaches the leader's inbox. The leader still reviews everything that matters; the drafts just arrive better.
Other lanes have their own products. Brev, built by people who spent a decade on OKR software at Ally and Microsoft, runs agents that update goals from the work and meetings already happening, draft the reviews, and flag execution drift early. Different lane, same pattern: encoded standards, structural distribution, measured results.
What stays with the human
The interesting line is what does not get handed over.
- Setting the standard: what "good" means is a leadership judgment. AI can apply a standard; it cannot decide what the organization should value.
- The hard conversation: performance, motivation, and trust are between people. No rubric covers them.
- Deciding what matters: which work gets done, in what order, for whom. Allocating attention remains the manager's job.
- Accountability: a scored draft is an input. Sign-off, and the consequences of it, belong to a person.
Operating rules for practicing MBAI
Five rules keep it honest.
- Encode the standard in writing: MBAI starts with an explicit rubric. If the standard lives only in the leader's head, nothing can be delegated, to AI or to a new hire.
- Delegate discrete tasks, not the function: choose bounded, repeatable work with a clear input and output. First-pass review qualifies. "Manage my team" does not.
- Keep the human in the loop by design: the AI's output arrives as a scored draft awaiting a decision, never as a decision already taken.
- Ground the AI in your own expertise, and respect the privacy line: standards, examples, and proven approaches make the output yours; open work channels, not private communications or inferred emotions, keep it legal and legitimate.
- Measure it: track scores over iterations and hours returned. A management practice that cannot be evidenced will not survive its first budget review.
Clayton
Ready to practise MBAI?
Start free, or schedule a free consultation with Clayton to see how it works for your team.
or email clayton@sagely.ltd