Executable architecture
Most enterprise architecture describes a future state. Ours is built to prove it. Every architectural decision connects to an executable specification, a working prototype and the evidence that it can succeed — independent of any vendor, and expressed in open standards (ISO 20022, BIAN) so you can change technology without rewriting the operating model. We don’t stop at PowerPoint; we build the evidence. This is the engineering discipline behind everything else on the site.
Architecture is the bridge from a proven decision to an executable system — with the cross-cutting disciplines holding at every layer.
The eight Intelligences produce the decision; architecture proves it can be built and governed; the reference runtime turns the specification into something that runs.
The Repository is not a document library. It is DTA’s executable knowledge graph: a connected model where the architecture is designed, challenged, evolved and evidenced before delivery begins.
It captures the relationships between business capabilities, journeys, the eight Intelligences, agents, data products, standards, regulations and decisions — so a change in one is traced to everything it touches.
Most consultancies accumulate documents. DTA accumulates executable knowledge.
300+ linked architectural concepts · hundreds of relationships · executable agent definitions · reusable patterns · architecture decision records.
Decisions are captured as Transformation Decision Records under the open MTDR standard (MIT, tool-neutral); architecture decision records are projections of those governed nodes.
An architect doesn’t ask “where’s the document?” They ask what a change touches. The knowledge graph answers in links, not folders:
The what is above. The how — the schemas, the mechanics, the model internals — is worked through under engagement.
Your knowledge should never be trapped inside a proprietary tool.
The Repository captures decisions, capabilities, agents, skills, evidence and transformation patterns in a structured, governed knowledge graph — then exports them to open formats such as OKF (Open Knowledge Format) so they can be moved, inspected, versioned and reused across platforms, with provenance, ownership and governance intact. We keep a richer internal model; you keep an open, portable asset. Owned, not rented.
Agentic processing only works when the journey, the decision model, the data products, the controls and the escalation paths are designed together. On their own, agents automate fragments of work that cannot safely be delegated. Design the agent last.
Eight agent roles, governed together — what each may do, on whose authority, with what evidence. See how reasoning is allocated in Cognitive and governed in Control.
Everyone claims BIAN, ISO 20022 and C4. The difference is that we link them into one model, end to end.
The Repository doesn’t stop at reference architecture. For a future banking journey, it connects the capabilities, data products, agent roles, controls, standards, evidence and delivery artefacts required to move from today’s operating model to tomorrow’s customer outcome.
Before a journey is funded, the architecture answers eight questions. Some agentic journeys are impossible until the architecture is fixed first.
Is the customer outcome measurable?
Are the required data products real?
Which decisions can be delegated?
Which decisions need human judgement?
What controls are embedded?
What evidence is produced?
What standards are touched?
What architecture must change first?
Stable, public views of the patterns we work with — each linked to a paper or a working demonstration. The deeper detail is worked through under engagement.
C3 — Capability, Context, Consent: our most-used reference architecture, beneath the eight Intelligences. Read the paper →
The reference architecture for governed human-machine action.
Reusable banking capability units — the unit a squad owns.
Evidence-grade data products beneath the engines.
One governed, agent-native platform carrying many propositions.
Resilience as a property of the architecture, not a bolt-on.
The executable blueprint between a reference architecture and a build.
How we apply the architecture in an engagement.
We publish concepts, outcomes, named roles and the shape of evidence. The detailed schemas, control internals and model mechanics are worked through under engagement — which keeps the public views honest and the methods protected.
Traditional enterprise architecture asks what should we build? We ask what evidence do we need before we decide to build it? Every transformation begins with a decision — and the architecture is the evidence that the decision is sound.
Bring us your most difficult transformation problem and we will show you how to prove it before delivery begins. The full model is walked through under engagement.
Tell us the proposition, architecture decision, or transformation problem you are facing. We will tell you, honestly, whether and how we can help.
Discuss a challengeLiving models and Field Notes on the future of banking. Published when the model changes, not on a schedule.