What is independent transition assurance?

A named expert, independent of the owner and independent of everyone delivering the change, accountable for measurable improvement in one dimension of the operating model. The role guides without overriding the owner's autonomy, validates the evidence, blocks subjective decisions and escalates when an owner and a governance body disagree. Nobody can credibly assure a transition they are also delivering, which is why the assuring party must not be the build partner, the offshore vendor or the systems integrator.

Balanced transformation, owned locally

Build strength,
not size.

Q: Our source code does not leave our network. Is that a problem?

No. Everything runs inside your own tenant or on your premises, air-gapped where required, with your own model if you prefer one, and nothing crosses the boundary. In defence, government, central banking and market infrastructure this is the entry condition, not a feature comparison.

Transformation run centrally pulls one lever at a time, and stalls. We put improvement in the hands of the people who own the work, balanced across every dimension, with independent evidence it is happening.

Start a conversation Read the argument

Designed and implemented inside two large regulated financial institutions, once across four hundred services and once department by department. The framework travels. The judgement about which dimension to pull is the work.

The problem

Some transformation belongs at the centre. Most of it does not, and the reason is arithmetic.

Cost out, work moved, vendors exited, legacy retired, a regulator satisfied, something back from AI. Almost every enterprise runs all of it the same way: one central programme, one target number, one office holding the levers.

A programme office holds twenty things.

You have four hundred services, each with its own starting position, regulatory load and backlog. It reaches the loudest few. Everyone else waits, and while they wait they optimise for delivery, because that is what they are measured on.

So move the ownership, not the programme.

The person who already owns the work owns its improvement, as part of the job. The centre holds guardrails and reporting, and does not execute. Every owner moves in parallel.

To be clear about what should stay central. Horizontal capabilities built once and inherited by everyone belong to a programme, and a programme runs them well: a new location stood up, an engineering standard, a control framework, a shared platform. Do those centrally. What does not survive being run centrally is the improvement of the work itself, because that is four hundred different starting positions, not one.

Then the second failure arrives.

Whatever gets pulled hardest moves alone. Cost comes out of people while engineering practice stays the same, so quality drops and the saving reverses. Work moves without knowledge transferring, so incidents arrive at month three to six, when fixes cost most. Vendors consolidate without technology simplifying, so the same complexity is bought at a better rate.

Every lever is defensible. The system still gets worse.

Transformation holds only when it is owned by the people who run the work, balanced across every dimension, and evidenced by someone independent of delivery.

The framework

Every owner declares a position on every dimension. Not just the one under pressure.

You cannot pull one without paying somewhere else. Move people to a cheaper location and delivery quality takes the strain. Cut run cost hard and resilience degrades quietly, until an audit or an incident.

Balanced target Actual position Paying for it

People. Vendors. Customer Experience. Security and Resilience. Investment and FinOps. Technology and Architecture. Engineering Excellence.

Seven positions, declared by the same person who owns the work, reviewed on the cadence of the business review that already exists rather than in a separate transformation governance. The chart is one owner's real position: two dimensions pulled hard, three paying for it, and a target nobody is tracking against.

What each one contains, and what is currently pressing on it, is set out in the next section.

This set is indicative, not fixed. The dimensions align to the corporate goals of the organisation, so the list changes between clients. Choosing it is part of the work.

Nothing counts as improvement unless it moves a measure declared before the work started.

The pressure now

AI is not an eighth dimension. It is a force on all seven.

An AI programme, an AI lead, an AI budget, running beside the operating model rather than inside it. That is the mistake this framework exists to prevent, and it is why most enterprises cannot show a return. Below: each dimension, what AI is actually doing to it, and the question its owner cannot yet answer.

People

Knowledge transfer compresses from months to weeks. The skill many teams were built to supply commoditises at the same time.

If capacity is what commoditises, what is my team for?

Vendors

The unit cost of a supplied engineer has not moved, so a single-digit discount does not add up.

What is my supplier actually charging me for now?

Customer Experience

Demand arrives already AI-generated. The volume shock comes from outside, before anything is adopted inside.

My inbound changed shape. My capacity did not. Who owns that?

Security and Resilience

An agent needs access, and access is where internal AI tooling fails, long before capability does.

Can I evidence what the agent saw, and who authorised it?

Investment and FinOps

AI for cost gets funded. AI for growth does not. Token spend is a new unit cost with no owner.

What is my cost per unit of AI-assisted work?

Technology and Architecture

The model is no longer the constraint. Legacy integration is, and that is where the benefit disappears.

Which parts of my estate can I actually point this at?

Engineering Excellence

The tools are already in engineers' hands. Knowing where it is safe to use them is not.

I gave my engineers the tools. Why has lead time not moved?

The constraint on AI value is not capability.It is permission and evidence.

What we sell

The Accountable Expert.

A named expert, independent of the owner, accountable for improvement in one dimension. Not a programme takeover. The role has already been built into two regulated institutions, housed differently in each, which is why it survives past a phase gate.

Ensures the work improves in a measurable, structured way
Guides without overriding the owner's autonomy
Validates the evidence and blocks subjective decisions
Escalates when the owner and a governance body disagree

Seven positions, held at once, by one person. That is the job.

Durable in a fast market

In 2024 this seat argued about CI/CD coverage. In 2026 it argues about which parts of the estate an agent can safely touch. The seat persists, the content refreshes, and you do not re-procure a firm every time the pressure moves.

Independence is now regulatory

Nobody can credibly assure a transition they are also delivering. When a supervisor asks how you knew the transition was safe rather than reported as safe, the answer has to come from somewhere.

The category already exists

A failing English council gets an Improvement Board: an independent chair appointed to assure recovery, who does not run it. Not a new idea. An established institution, applied where it currently does not exist.

Where you are

Four situations bring people here.

"Our transformation reached a few areas and stalled everywhere else."

Central, under-delivering, and you already know it. The arithmetic is the opening.

Best mid-flight

"We deployed AI. I cannot show a return and I am being asked why."

Because it runs beside the operating model, not inside it. The seven dimensions are the diagnostic.

Best with a CIO, CTOO or AI sponsor

"Concentration in one delivery geography is now a board risk question."

Most boards call it material. Few have quantified it. The answer is only actionable per capability, not per enterprise.

Best with a CISO, CRO or risk-led CIO

"My managed-service renewal is coming and the numbers no longer work."

Time and materials is under terminal pressure. A single-digit discount does not reflect what has actually happened to delivery.

Best with a large managed-service estate

In practice

Nine situations we are usually called into.

Each is voiced by a client in their own words, long before anyone mentions a method. Eight are engagements. The ninth decides whether any of them can happen at all.

Each one has its own note, here, with what actually goes wrong, what we do about it and the questions we are always asked.

Knowledge and people

"The people who understand this are leaving."

Two people understand a system and one retires in March. Every conventional answer needs the person walking out of the door. We evidence what genuinely depends on fewer than three people, then design the succession, the transfer and the run model around what we find. Dual-running is the line item most likely to break an offshore business case; this is what closes it on schedule.

Comes up with a named retirement date, a capability-centre move, or a vendor exit

Read the full note on key person risk and knowledge transfer

Standing up a capability centre

"The board approved the centre. Nobody has said what it is for."

A location decision usually arrives as a headcount number and a cost target. What is missing is the design: which capabilities move and in what sequence, what the receiving organisation actually looks like, who owns what on day one, and what has to be true before anything is handed over. In the absence of that, the centre gets designed around supplying engineering capacity, at precisely the moment engineering capacity stops being the scarce input.

We classify the capabilities before anything moves, so where work goes is the output of a judgement rather than a policy. Then we design the receiving model around what does not commoditise, and set the evidence gates the transfer has to pass before the saving is recognised.

Comes up when a centre is approved but not designed, at a second-wave scale-up, or when an existing centre is delivering capacity and not capability

Read the full note on capability centre design

Regulatory and risk

"We cannot prove this to the regulator."

The second line of defence has no independent evidence, so it ends up asking the first line it is supposed to be challenging. Every audit rebuilds the same picture by hand and it is stale on delivery. We locate where control logic actually sits against where policy says it sits, and leave a record that refreshes rather than being rebuilt.

Comes up after a regulator finding, before a scheduled audit, or against a dated obligation such as DORA

Read the full note on regulatory evidence and audit

Portfolio and strategic decisions

"We are committing money against a picture nobody trusts."

At the moment of largest commitment, the estate picture usually comes from the vendor who profits from the renewal, or from the integrator sizing the programme it intends to bid for. The configuration management database is wrong and the diagrams are three reorganisations old. We build the picture independently, from the systems themselves, and separate what is genuinely locked in from what is portable.

Comes up at a contract renewal, an end-of-life platform, or a consolidation mandate

Read the full note on independent estate assessment

Mergers, carve-outs and corporate events

"Forty-five days, two million lines, and one person spot-checking it at night."

In a software deal the software is the largest asset and the only one nobody reads. Every other diligence stream has been industrialised. This one is still sampling, and buyers price the resulting uncertainty as a discount without ever calling it by its name. We read the estate inside the deal window: what is owned against what is licensed, key-person concentration before the earn-out, and how entangled the target is with a business that is staying behind.

Comes up in exclusivity with a diligence window, an approaching exit, or an announced carve-out

Read the full note on technology due diligence

Modernisation execution

"It was quoted at three years, so we deferred it again."

A modernisation makes thousands of decisions about what survives, and today almost none of them are made deliberately. A payment capped at nine invoices because a print band was a fixed width is not a policy. Nobody chose nine. It is rarely forty-year-old COBOL either; more often a 2018 system drowning in libraries nobody can patch. We retire the dead logic before the programme commits to carrying it.

Comes up at end of life, when a vendor forces a move, or on a stalled programme that has already spent money

Read the full note on legacy modernisation

Delivery and change economics

"Our estimates are planning poker."

Clients pay for not understanding their own systems every single day, and it never appears as a line item. Every release is a gamble, so the team asks whoever has been there longest. Engineering dashboards report a high change failure rate and cannot say why, because they do not read the system. This is the one situation where the client already holds the baseline, sitting in their own incident and release records.

Comes up after a serious incident, a missed commitment, or at a vendor rate review

Read the full note on delivery and change economics

AI enablement

"The board told me to do AI. I have activity, not a programme."

AI works on code written last month and fails on the estate that runs the business, because coding agents sample and cannot tell a business rule from an accident of a record layout. So the experiments are all greenfield and the mandate produces motion rather than change. We establish which parts of the estate are legible enough to be safe to point an agent at, and what has to be true before the answer is yes.

Comes up with a board AI mandate carrying a date, a new Head of AI, or a rollout that cannot show value

Read the full note on AI enablement

Sovereignty and deployment

"Our source code does not leave our network. That is not negotiable."

For some organisations the first question is not what the work is worth, but whether they are permitted to do it at all. This is a qualifying gate rather than an engagement. Everything runs inside your own tenant or on your premises, air-gapped where required, with your own model if you prefer one, and nothing crosses the boundary. In defence, government, central banking and market infrastructure this is the entry condition, not a feature comparison.

Comes up when a security review is already blocking a purchase, under a sovereignty policy, or after a vendor has been rejected on data egress

Read the full note on sovereign and in-tenant deployment

The offer

Fixed outcomes behind gates, then a retained independent role.

Deliberately not a programme, because the argument is that programmes do not scale. Each module is a fixed deliverable behind a go or no-go. The baseline earns the design.

The baseline · 3 to 5 weeks

Compass

Structured assessment across the dimensions with the owners themselves, not a document review. Produces a current position, a comparable view across owners, and the two or three places you are actually exposed.

Current position · exposure findings

The target · 4 to 6 weeks

Blueprint

Target position per dimension per owner, the operating model that follows, and the sequencing judgement: which dimension moves first, and what it will cost elsewhere.

Target positions · sequencing

The enablement · 3 to 4 weeks

Academy

Equip the owners to run it themselves. Playbooks, target-setting facilitation, review cadence. This decides whether anything survives our departure, and it is the step most firms leave out because it ends their engagement.

Playbooks · owner workshops

The evidence · throughout

Vault

The measurement frame, the dated record of position and movement, and the audit trail that answers a supervisor asking how you knew a transition was safe rather than reported as safe.

Dated record · audit trail

The retained role · 12 months

Watch

The Accountable Expert, retained and independent, priced per dimension. Monitors, challenges, validates, escalates, reports. Where independence stops being a claim.

Per dimension · monthly review

Supporting

Canvas

The AI tooling, used where an estate has to be made visible before it can be safely moved, modernised or automated. Support to the method, never the headline.

Estate intelligence

The scope boundary on Watch protects both sides. We monitor, report, hold delivery parties to account and escalate. When a vendor or an internal team slips, we surface it, quantify it and define what good looks like. Remediating it is their job, or a separately scoped package. The retainer does not silently absorb someone else's execution labour.

Why us

Four things that are hard to imitate.

We designed it. We did not read about it.

Built with named accountable experts across every dimension, then implemented at two regulated institutions with different regulators and different starting positions. The differences between them are documented and deliberate. Harder to imitate than a methodology deck.

We are independent of everyone who delivers.

Not the build partner, not the offshore vendor, not the systems integrator, and we do not compete with them. Any firm that both delivers the change and assures it has a problem the moment a regulator asks.

We have worked both ends of the delivery line.

Between an EMEA or US headquarters and a delivery centre elsewhere sits a gap on escalation, seniority, ownership and what "done" actually means. We have built and run teams on both sides of it, and we translate between the room that sets the mandate and the floor that has to meet it. For most firms that gap is a hidden, recurring failure mode.

We build with the technology, not about it.

Strategic product advisor to an AI product that reads enterprise codebases. The AI claims here are grounded in building, not in a vendor briefing.

Where the tooling sits, and where it stops.

Technology and Engineering are the two dimensions where a stated position is least reliable, because nobody can describe an estate they have not read. We apply AI analysis to the codebase, the schemas and the pipelines, and ground the target position in fact. It runs in-tenant. Source code does not leave and is not used to train anything.

Where it stops, stated deliberately. Strongest on codebases it can read. Weaker on closed commercial software where only configuration is visible, and on obscure stacks, where automated insight has to be supplemented by human capture. A tool pitched as having no limits is the trap this whole argument accuses others of.

How it is delivered

A team built to the engagement, not a bench that has to be fed.

Accountable experts are drawn per dimension rather than from a fixed bench, which is the same design used inside the client organisations themselves.

Framework and operating-model design
Estate and knowledge visibility
Technical and vendor depth, from inside the detail
Capability-centre and vendor-exit sequencing
Stakeholder management, because ownership is political

Savings, productivity and risk figures from prior programmes are shared once an NDA is in place. The method is shareable. The outcomes are not.

Start a conversation

Tell us what you are trying to land.

No deck follows automatically. A short conversation about your situation and whether an independent accountable role is the right instrument. If it is not, we will say so.

Name

Work email

Organisation

Role

Which door are you coming through

What is the situation

We use what you send only to reply. No mailing list, and we do not pass it on.

What happens next

A reply within two working days, from a principal
A short conversation. No deck, no workshop, no pipeline process
If there is a fit, a mutual NDA. That is where the numbers and the named references live
If there is not, a straight answer and a pointer to who is better placed

Build strength,not size.