Transformation Intelligence
Transformation Intelligence is the capability to understand the organisation: why it is as it is, what is changing, and what remains unknown. It rests on a governed, computable model of the organisation, one that can be queried, projected and reconstructed to understand and reason about change.
Every function has a model. None of them is wrong. None of them is the organisation.
Knows what everything costs. Cannot tell you which authority depends on a role.
Knows who is employed. Cannot tell you which jobs an agent could legitimately perform.
Knows which systems exist. Cannot tell you why a platform exists, or which propositions disappear if it is removed.
Knows which controls are in place. Cannot tell you which consequence each one is actually protecting.
Knows how the processes run. Cannot tell you which of them the customer's job still requires.
Knows the outcomes it wants. Cannot tell you what the organisation would have to become to reach them.
Each account is coherent on its own. Nobody receives an account of the conflicts between them: where one function's optimisation destroys another's value, where accountability exists without control, where a capability the outcome needs already exists but cannot be reached. That cost has a name, institutional latency: the time an organisation takes to understand what has changed, establish who may act, decide, and act coherently. It is why organisations lose to circumstances they saw coming.
Start with the job, never the structure. A customer still needs to make a payment whether the participant performing the work is a branch colleague, an operations team, a core application, a software service or an agent. Jobs are stable. Everything arranged around them changes.
Purpose and value justify jobs. Jobs require skills and authority. Participants are selected to perform them. Every action leaves evidence.
Your organisation today is the residue of consequential decisions. It was not designed once; it accumulated. Build or buy, centralise or distribute, human or agent. Outcomes were kept. The reasoning that produced them rarely was, and every one of those decisions carries the same five questions.
What did we choose, and what did we reject?
What outcome justified the choice? Value does not belong to the transformation. It belongs to the organisation and the customers it serves.
Who was entitled to cause the consequence? Access tells you what someone can reach. Authority tells you what consequences they are entitled to cause.
What authority, obligations and constraints survived into execution?
Can we show what actually happened? A runtime log shows what happened. A decision record shows why the organisation believed it was legitimate to let it happen.
These are not governance artefacts surrounding a transformation. They are part of the organisation, and they can be recorded, connected and queried like the rest of it.
Do not start with where to deploy AI. Start with the jobs the organisation needs performed, the skills each requires, the authority that must be exercised and the evidence that must survive. Only then choose the participant. Human, agent, service, software or supplier is a sourcing and assurance decision, taken after the job is understood and reversible when it changes.
The job is invariant. The participant is a selection.
Could your organisation say today why a particular job is performed by a person rather than by software, an agent or a supplier?
You have systems that compute your money, customers, risk, workforce and technology. But the organisation itself cannot yet be computed. Underneath Transformation Intelligence sits a model that makes it computable: not as a data model or a system of record, but as governed statements of what the organisation has established, what may be computed from them, and what should hold instead.
Records let an organisation know itself. Projections let it see itself differently. Reconstructions let it question whether it should remain that way.
A record is not truth. It carries the standing on which it was established, and later evidence can contradict it. A projection never becomes a record by being useful. A reconstruction starts from the job and the intended outcome, keeps the genuine invariants and constraints, and refuses to treat today's process, platform, role, participant or boundary as the baseline for what comes next.
See where the organisation's work converges, and how DTA makes it computable: Platform
Ask how a proposition actually works and follow the answer through the organisation: the jobs it requires, who or what performs them, the authority they exercise, who remains accountable, and the evidence behind each step. Then change something and watch what else changes.
How many systems would you have to interrogate to answer the same question in your organisation today?
Apply a decision and the organisation is worked out again from the result. Consequences propagate by function, and the model says which ones it cannot establish rather than filling the gap with a number.
Cost. Which work is created, moved or retired, and what that does to the cost of operating.
Skills and roles. Which skills become scarce or redundant, and which participants change.
Platforms and services. Which of them the work still converges on, and which lose their justification.
Authority and controls. Which authorities move, expire or no longer fit, and which controls follow.
Propositions and value. Which customer jobs are touched, and whether the value claimed still stands.
Accountability. Where the obligation to answer moves, and where it was never established.
Where the model holds no evidence, it says so. An unknown stays unknown. It is never quietly converted into zero.
A slide deck can describe the organisation. A computable organisation can tell you what your decision changes.
One set of Records answers differently depending on who is asking. Each answer is a projection: computed for a purpose, carrying its assumptions and provenance, never promoted into a fact.
An organisation view: what the organisation must do, who performs it, and who answers for it.
A value projection: what value each part of the work protects or creates, and what a decision does to it.
A workforce projection: which jobs exist, which skills they need, and which participants could perform them.
An authority view: who may cause which consequence, within what bounds, and where authority is exercised without a chain back to its source.
A capability view: where work converges on shared platforms and services, and what depends on each of them.
An evidence position: what the organisation can establish about a past action, and where it cannot.
Replace a person with an agent and, underneath, nothing changes: job, value, authority, accountability and evidence are exactly what they were. What has gone is the informal judgement that used to fill the gaps.
An agent being technically able to execute something establishes nothing about whether the organisation authorised its consequence. A tool can be exposed. A token can identify the caller. A policy engine can allow access. None of those establishes why this participant is entitled to cause this organisational consequence. Behaviour never establishes authority. Where software starts to act on a customer's behalf, the same questions apply: agentic payments.
We did not make the organisation computable for AI. We made it governable. AI simply makes the value of that capability, and its absence, impossible to ignore.
Two capabilities over one model of the organisation.
Order matters here. A consequence occurs. Evidence establishes what occurred. Only then does the organisation's record of itself change. A loop that moved straight from decision to changed state would let the organisation believe its own plans.
Underneath sit a small number of governed classes of Record, each with its own grammar. Two are open standards you can inspect today.
MTDR is one open recording grammar for one class of Record: consequential decisions. A Transformation Decision Record captures enough judgement to defend, execute, monitor, challenge and later supersede a decision. It is MIT-licensed and tool-neutral, and works alongside your architecture decision records, your wiki and your governance platform. Inspect the standard or read what a decision record is, and is not.
A Value Record holds a commitment to value and its settlement as one record, so that expected, realised and written-off value can be reconciled against the work that was meant to produce it. Published beside the decision record, in the same repository.
Who may cause which consequence, what the organisation owes, and what survives into execution are recorded as classes of their own, under the same discipline: standing, provenance, evidence and supersession, never editing.
Client records and the open standards survive without us. DTA uses the governed model as its advisory instrument, not as a product you have to buy.
We will show you what your existing models cannot tell you about its consequences.
We do not sell a framework. We do not launch a programme about a problem. We do not ask you to replace what already works. We solve the problem in front of you, then earn the next one.