Our Approach: Building the Organisation of the Future

From pilots to P&L

Two years in, most banks have a Centre of Excellence, a stack of pilots, and copilots that made people feel faster without moving the numbers. Value comes from going vertical: one journey, mapped end to end, with the work that carries the cost and the risk governed.

The deciding variable isn't budget — it's a board that will answer for what an agent does. The headline AI commitments are already public; the mid-tier question is whether you can move with that conviction on a fraction of it. We give you the architecture, the controls and the proof to do exactly that — with a way in that doesn't require betting the bank.

The principles we work to

Transform the bank without losing the bank. Done right, the people become a lever for growth, pointed at the propositions and markets they never had capacity to chase. And we will tell you when not to automate: a customer making a claim, or asking about their care, needs a person. Leaving no customer behind is a design rule, not a slogan.

Human-centred
Every change is judged by the outcome it creates for a real person.
Evolution over replacement
Run the old and new worlds together and migrate safely. Do not rip and replace.
Capability before technology
Build human and organisational capability first. Technology serves it.
Human-in-the-loop
Agents prepare and propose. An accountable human authorises material action.
Consumer outcomes
Design and evidence good outcomes for customers from the start, not as a post-launch audit.
Preservation before optimisation
Retain knowledge, evidence and institutional memory before optimising.
Continuous control
Accountability and control are properties of the system, evidenced continuously.
Standards and interoperability
Build on open standards so the work is portable and vendor-neutral.

Architecture and strategy come first

Before a line of code, we work out what proposition creates value, how it makes money, and how the organisation must be shaped to deliver it. That decision spans five lenses at once.

Business architecture Domain architecture Technology architecture Enterprise architecture Strategy & proposition design

Proposition, operating model and architecture are one decision. From it we set a high-level vision and a roadmap, pitched roughly two years out, that moves people, process and technology in step. Get the order wrong and architecture choices arrive too late to matter.

This page is the operating-model leg of one capability: the target architecture (what good looks like, proven), the operating model to build it, and the capability intelligence to staff it. Most banks bolt an operating model onto a platform they chose before the hard parts were proven. We define the architecture first and build the model that delivers it.

This is the evolution method in practice. We read the work through the Eight Intelligences, design against reusable reference architectures, prove the hard parts in a working model, and run the old and new worlds together rather than betting the bank on a single cutover.

An opinionated model, on industry-proven frameworks

We don't reinvent what works. We assemble proven industry frameworks into an opinionated model, then customise it to your organisation.

BIAN SAFe TOGAF (ABBs) SFIA (people layer) DBBs (our term)

BIAN, SAFe, TOGAF and SFIA are established standards. SFIA, the Skills Framework for the Information Age, grounds the people layer: every role is a SFIA skills profile, so capability is explicit and portable. Domain Building Blocks (DBBs) are our own term, coined to sit alongside TOGAF's Architecture Building Blocks. An ABB is a reusable technical capability; a DBB is a reusable banking capability, tailored to you.

How it all fits together

One sequence. Senior discovery defines the value and target organisation. A Glass Tube pilot proves what matters, as a light sandbox proof or a full organisational slice. Squads then scale the DBBs on a Capability·Context·Consent architecture.

flowchart TB
  OA["OWNED AUTONOMY: accountable and sovereign by construction"]
  DISC["Discover: DTA leadership, senior architects and domain specialists"]
  PILOT["Pilot, the Glass Tube: light sandbox proof or full organisational slice"]
  subgraph C3["C3 architecture spine"]
    CAP["Capability: Domain Building Blocks"]
    CTX["Context: shared semantic layer and measurement"]
    CON["Consent: authority, policy, evidence, agentic gateway"]
  end
  subgraph VS["Value stream: proposition and P and L"]
    SQ["The squad: Product Owner, Forward Deployed Engineer, 2 Engineers, Data Engineer"]
  end
  BANK["Selected Glass Tube and scale enablers: Agile and agentic coaches, Risk, Audit, Finance, HR, Procurement, Operations"]
  FND["Foundation (shared): substrate, agentic harness, paved-road controls"]
  BACK["Backbone and leadership: accountable sponsor (SMCR), enterprise architect, sequencing lead"]
  DISC --> PILOT
  PILOT --> SQ
  BANK --> PILOT
  CON --> CAP
  C3 --> OA
  SQ --> CAP
  VS --> C3
  BANK --> SQ
  FND --> SQ
  BACK --> VS

The deeper operating-model detail (coaching, the leadership shift, the transformation backbone, and landing in a not-yet-ready environment) is on the operating model page.

Your control plane, not a rented one

Every agentic bank needs a control plane. What decides sovereignty is who operates it.

The Consent plane in that diagram is the bank's control plane: the agentic gateway, the authority registry and the evidence store, operated by the bank itself. Settlement rails plug into it as versioned building blocks, Faster Payments and ISO 20022 today, with SEPA Instant, RTGS and future central-bank rails on the same contracts. Identity and credential schemes plug in the same way, and engines for fraud, financial crime and risk sit on top as thin layers over governed data. Because the bank operates the plane, it decides what plugs in, and can swap or revoke it. Rent that plane from a vendor or a protocol and you have rented your sovereignty.

Faster Payments and ISO 20022 messaging run in Accountable Bank, our demonstration. Multi-rail and future-rail support is a design blueprint.

Squads, Domain Building Blocks and value streams

A squad is useful before it becomes an operating model. Clients can consume one senior agentic squad for a bounded outcome, or group durable squads into value streams as their ambition grows.

Domain Building Blocks

A DBB packages a meaningful banking capability (journey, policy, controls, evidence and execution) that is reused across propositions instead of copied. It is our term, alongside TOGAF's ABBs, and the Capability plane of our C3 architecture. Read the paper →

Squads own the DBBs

Each squad owns one or more Domain Building Blocks end to end. Small enough to move fast, senior enough to be accountable for what they ship.

Value streams (SAFe)

Squads group into value streams aligned to a proposition and its P&L. SAFe gives you coordination across squads without re-centralising delivery.

Inside a squad

Deliberately small. One squad, five people, accountable for its Domain Building Blocks.

1Product Owner

1Forward Deployed Engineer

2Engineers

1Data Engineer

Every role is defined against SFIA, the global skills standard, so capability is explicit and assessable. The Product Owner owns product management, requirements and stakeholder relationships, and is the single engagement point to the business. The Forward Deployed Engineer (FDE) owns solution architecture, systems integration and consultancy, a forward-deployed role that integrates into your estate. Engineers own programming, testing and deployment; the Data Engineer owns data, evidence and measurement. (Skill names shown; the full SFIA profiles with levels of responsibility are part of our IP. Built by a contributor to the SFIA framework.)

It flexes. AI is blurring role boundaries fast, and SFIA lets us name which skill AI has absorbed rather than argue about job titles. A light Glass Tube begins with the Product Owner and FDE as a forward-deployed pair; a fuller Glass Tube adds the roles needed to run the domain end to end. At scale, squads bring paved-road controls, integrate through the FDE and run on fixed-capacity funding: cost is the constant, scope flexes, and each squad owns its cloud and AI run-rate.

The old squad model

The two-pizza idea expanded into large delivery groups
Work split across layers, hand-offs and offshore rate cards
Capacity measured in people and hours
Knowledge dispersed across suppliers

The agentic squad

Small, senior and AI-amplified
Owns a bounded outcome or Domain Building Block
Capacity measured by evidence and shipped value
Knowledge and capability stay with the client

The whole organisation reorganises around the squads

Banks are cutting costs everywhere. The model's value is showing what to protect and invest in, because the squad framework only works when the whole business moves with it.

Our job is to show what's genuinely important: the capabilities, controls and architecture decisions worth protecting, whether you run an established bank, are standing up a new sub-division, or are a founder starting from scratch.

Risk, finance, HR, regulatory affairs, compliance and operations each have to align to how value is now delivered and evidenced. For an established organisation the risk is fragmentation: these functions siloed and duplicated per product. For a new sub-division or a startup it is the opposite, under-building the real bank and then discovering its weight in production.

A real bank is heavy: multi-currency, multi-country, multi-regulator. Even an ambitious "zero-ops" model still needs a substantial, accountable business wrapped around it. Automation reduces effort; it does not remove the obligation.

From

Functional silos
Projects & programmes
Status reporting
Duplicated stacks per product

To

Value streams
Durable squads
Accountable autonomy
Evidence over status

Getting there from where you are

Most of our work is whole-organisation transformation for established banks and FinTechs, and the same model gives teams building something new, a sub-division or a startup, a way to start right.

If you already run a bank or FinTech

The hard part is sequencing change while keeping the lights on. We start with senior-led architecture and strategy, pick a beachhead proposition, then run a Glass Tube pilot at the depth the decision requires. It may begin with a Product Owner and FDE building a sandbox prototype, or extend through the first domain with a squad and the relevant business functions. Shadow integration proves value alongside the existing estate without re-platforming. Then you expand block by block, migrate the business functions in step, and decommission the old path only once the new one is proven.

If you're building something new: a sub-division or a startup

Here the hard part is the reverse: building enough bank without a cost base you can't carry. We start from the proposition, the P&L and the licence perimeter, then use the Glass Tube to prove the solution and as much of the operating model as the decision requires. From there, the first durable squad forms around the smallest set of DBBs that makes the proposition real, alongside the minimum credible bank a regulator and customer require.

Deploy a squad. Prove a domain. Transform the bank.

Three buying motions, each useful on its own. Start with the outcome and choose only the depth of change it needs.

Deploy an agentic squad

The simplest route: a senior engineer, a managed squad or a fractional lead focused on a bounded outcome. No transformation programme required.

Run a Glass Tube

Choose the depth of proof, from a PO/FDE-led sandbox prototype with selected enterprise services to a full domain and organisational slice.

Transform and scale

Expand into durable DBB squads, value streams and whole-organisation change across funding, people, sourcing, controls and production architecture.

From senior discovery to organisational scale

The Glass Tube grows to match the decision it must support.

Founder-led discovery Glass Tube pilot Light or full-depth proof Durable DBB squads Evidence-led growth

Discovery produces the proposition and Capability Blueprint: the value, domains, human and AI-assisted capability, evidence and gaps. It then proves the proposition at the required depth. Its lightest form is a PO/FDE-led sandbox prototype; its fullest form runs a domain end to end with the first squad, coaches, controls and functional specialists. Scale follows the evidence. See Capability Intelligence.

The Glass Tube

Our pilot is the Glass Tube: a transparent vertical slice sized to the uncertainty and the investment decision.

A light Glass Tube proves the solution shape, controls and integration path through a working prototype in a sandbox. It can draw on a few enterprise-service specialists (platform, security, data, integration or service management) where the proof crosses their boundaries. A full-depth Glass Tube runs one domain end to end, adding the first squad, coaches, funding, roles, sourcing, risk and audit.

Decide a proposition. We help you prove it.

Choose a proposition, say the Family Wealth Hub, and we help you build a spike for it. Standalone, or shadow-integrated alongside your existing bank so you prove value without re-platforming.

Accountable Bank, our demonstration, shows the technology is genuinely possible, and that it connects directly to business value in an agentic way: agents that observe, prepare, check and execute within explicit authority and human approval. Four propositions are interactive demonstrations; Franchise Network is a blueprint. They run on synthetic data: a working proving ground, production-grade in design and built for proving propositions, deliberately not a production system that runs with real customers.

Explore the demonstration Prove your proposition

Build the organisation of the future, on purpose.

From pilots to P&L

The principles we work to

Human-centred

Evolution over replacement

Capability before technology

Human-in-the-loop

Consumer outcomes

Preservation before optimisation

Continuous control

Standards and interoperability

Architecture and strategy come first

An opinionated model, on industry-proven frameworks

How it all fits together

Your control plane, not a rented one