The manifesto · a thesis for the next service economy

The Human Operating Model.

Software will define the service. AI will run it. And human beings will be the reason customers choose yours. One argument about what replaces shift left, why the ticket is evidence of failure, and where people create value nothing else can, drawn from the operating floor of a £550M services business.

I · Software defines the service II · Humans differentiate it III · Shift right

01 · The claim

The industry's one big idea has expired.

For thirty years the services industry has run on a single big idea. Shift left. Push work towards cheaper resolution, automate more, deflect more, keep humans away from demand. It was never foolish. In a world where every fix cost you an hour of somebody's time, it was sensible arithmetic. The problem is what happened next. We took an optimisation and made it doctrine, and once something's doctrine nobody asks whether it's still the right answer. They just ask which version of it we're on.

Then AI turned up. And it didn't beat shift left at its own game, it made the game pointless. When software can spot the problem, decide what to do and fix it before anyone notices, there's no left left to shift to. The cost question is busy answering itself.

Which leaves the question shift left was never designed to answer: where do human beings create value that nothing else can?

Not human as fallback. Human as differentiator.

That's the Human Operating Model. Not a framework you can get certified in. An argument about how services should be designed, sold, delivered and led when software does the running and people do the differentiating.

02 · Shift right

From the cost curve to the value curve.

Shift left moved work down the cost curve. Shift right moves people up the value curve, putting human attention where the stakes, the feelings and the consequences actually are. It isn't another workflow diagram. It's an experience architecture. Four modes. Not tiers, modes. Nobody has ever been thrilled to reach tier three.

Experience modesAI → human

i · AI

Self-heal by default

The problem resolves itself before the user knows it existed. AI operating within policy. Invisible, frictionless. The best service experience is the one nobody notices.

ii · AI +

Guided self-help

The user needs to act, but they are not alone. Intelligence and the user working together to clarify intent and find resolution. Not a FAQ page. An intelligent conversation.

iii · Human + AI

Assisted resolution

Human and AI together. The human validates, steers and exercises judgement; the AI carries the administration. Where complex and sensitive demand lives, because a human in the loop adds something automation cannot.

iv · Human

White-glove moments

Intentionally, fully human. Not because the automation ran out, because the situation demands it. Restoring confidence, exercising real judgement, showing genuine accountability. That is not a cost to be minimised. That is the value.

Underneath the four modes sits something the old model never had. A service sensor layer. Signals from users, from systems, from the business itself. Sensing, interpreting, deciding, changing, all without waiting for a ticket to turn up and announce that something's already gone wrong.

Service sensor layerSignal → change

Source 01

Signals from users

Source 02

Signals from systems

Source 03

Signals from the business

Signal→Insight→Decision→Change

By the time a ticket appears, the experience has already failed.

We've spent twenty years getting better at what happens after the ticket arrives. The next-generation service organisation gets better at not needing the ticket.

03 · Software-defined services

The operating model becomes a product.

None of this survives as a slogan painted over a manual operation. Shift right is only honest when the routine genuinely goes away, and that means the service itself has to be defined in software. Provisioned in code. Assured by telemetry. Resolved by agents. Improved the way products get improved, which is continuously, rather than the way documents get reviewed, which is annually and by someone who stopped reading on page two.

Software-defined service fabricTrust controls

Intent

Policy

Telemetry

Agents

Knowledge

Trust controls · Confidence · Audit · Rollback

This is the part the industry keeps missing about AI. AI lands or stalls in the operating model. You can't layer intelligence onto an unintelligent structure and expect a transformation. Most operating models were built for a different problem. They were built for tickets rather than flow, frameworks rather than fluency, process compliance rather than customer confidence. No amount of platform spend fixes that, though a great many people are going to try.

When the operating model is software, it stops being a diagram on a wall and becomes a product you ship. Services stop being org charts with SLAs attached, and start being software with judgement attached.

04 · The human differentiator

The floor is rising. The ceiling is human.

AI is levelling everyone. Every provider is now running on the same models, which means we're all about to get equally good at the boring bits. The average email. The average deck. The average resolution. All of it available to anybody, at roughly the same price, by about teatime. Competence used to be the pitch. It's now the floor, and the floor is free.

What's left is the ceiling. Trust, taste, language, the feeling of being in good hands. And people don't experience services rationally, whatever the satisfaction survey tells you. They say they want speed. They mean they want certainty. They say they want a response. They mean they want to believe somebody competent is holding it. It's rarely the waiting that upsets people. It's the not knowing. Clarity is part of the service. Expectation design is part of the service. None of that is soft. It's the most hard-headed thing on this page.

Feelings are not a side effect of service delivery. They are the product.

This is where our industry should be learning from the people who worked out how perception works long before we did. Behavioural science. Marketing. Hospitality. Every service business that understood how something feels is what it is. The next great idea for IT services won't come from IT services. It never has.

The differentiator of service is human. Not human as a cost line that automation failed to remove. Human as the one advantage nobody can download.

05 · The principles

Ten principles of the Human Operating Model.

Software defines the service.

If a machine can sense it, decide it or resolve it, it should, invisibly, within policy, before anyone needs to ask.

Humans differentiate it.

Deploy people where they change how the moment feels, not where the process happened to run out.

iii

Shift right, deliberately.

Four modes, not three tiers. Human attention belongs at the moments of dread, judgement and consequence, not the password resets.

Feelings are the product.

Until services are designed for how people actually think, feel and behave, they will keep working on paper and failing in practice.

Sell confidence, not time.

Nobody buys hours. They buy the feeling that the outcome is safe in your hands. Price it, design it, deliver it.

Design the uncertainty out.

It is rarely the waiting; it is the not knowing. Clarity, ownership and expectation design are operational disciplines, not soft extras.

vii

Language is architecture.

If customers need a glossary to understand your catalogue, the problem is not their vocabulary. Every word is a design decision.

viii

The floor is not where you compete.

AI raises everyone's floor at the same speed. Strategy built on efficiency alone is a detailed plan to become average, faster.

Steal from outside.

The next great idea for services will come from behavioural science, marketing, design and hospitality, not from another framework revision.

Rigour on the question, not the answer.

The leaders who matter next are uncertainty experts, not certainty performers. They ask why we do it this way, and listen.

06 · Where this comes from

Working notes, not commentary.

This thesis wasn't assembled on a whiteboard. I'm testing it daily, inside a £550M services portfolio, across five service functions and three hundred people, against thousands of customer environments and an ecosystem of more than eighty delivery partners. And it carries the scar tissue of both sides of the table. I've spent two decades running managed services at scale, and years as the customer holding providers to account, which teaches you things about your own industry you would rather not know.

It's been argued in public, on the main stage at SITS 26 and in keynotes at SDI Spark 26, itSMF UK and PropelX London. And in private, in the boardrooms where AI investment actually gets decided. It'll keep changing as the evidence does. That's the point.

£550M portfolio ~300 people 80+ partners NPS 68 in year one Both sides of the table

07 · From argument to Monday

Agreeing is the easy part.

Everything above is a position. The hard part is the Monday after, when you open your own demand data and find out how much of it you created yourself, how little of your elapsed time the work is actually moving, and what the first of those is costing you a year. Almost no service report contains those three numbers. They're usually the ones that change what a board decides.

Shift Right is this argument turned into six working modules that produce those numbers from your operation, and assemble them into one operating model with the diagnosis, the first ninety days and a board case built from your own figures. Same thesis, made accountable to your evidence rather than mine.

See a complete worked example Or read it as a PDF What Shift Right is

The manifesto, the essays, the radar and the benchmark stay free. This is for when you want to do something about it.

Asked plainly

The terms, in plain words.

What is the Human Operating Model? A thesis for how IT services should be designed now that software can run them, written by Mark Boyer and tested daily inside a £550M services portfolio. The short form: software defines the service, humans differentiate it, shift right. Provisioning, assurance and routine resolution become code you ship. What remains, trust, judgement, language, confidence, is what customers actually choose you for. The differentiator of service is human.

What does "shift right" mean in IT services? The deliberate redeployment of human effort to the moments where it changes the outcome: dread, judgement and consequence. Software absorbs the routine. People stop being the expensive fallback for whatever the tooling cannot do, and become the reason the service is chosen. Human as differentiator, not human as fallback.

Is shift right the opposite of shift left? Yes, and on purpose. Shift left was the right answer to an old constraint: human effort was scarce and expensive, so you engineered it out. That constraint has expired. When AI gives every provider the same routine competence at the same price, engineering humans out of your service is a detailed plan to become average. Shift right treats human attention as the scarce thing worth designing around.

Take the argument

Now put it in a room.

The manifesto is the argument written down. The keynote is the argument made. Hear it live, read the essays behind it, or get the working notes as they happen.

Book Mark to speak Read the essays

Shift Right: first access · Get the Dispatch