See before choosing.
Technology comes after understanding the flow, never before.
The 2dAI method · 05 validation gates
From a real problemto a reliable system.
A method that is simple to follow and demanding at every stage. We move forward because concrete evidence supports the next step, not simply because a schedule says so.
Share my problemThe principle
Evidence drives the project.
Validate before expanding.
Every cycle must produce an observable result before the scope grows.
Document before delivery.
A system is only robust if your business can understand, operate and take ownership of it.
The full protocol
Five stages. Five explicit decisions.
SHARED REALITY
Understand
See the work as it is actually done.
We begin with the real flow: people, tools, data, decisions and exceptions. The goal is not yet to choose a technology, but to identify precisely where the system is losing time, reliability or capacity.
What we do
- Observe users and handovers
- Inventory tools, data, access and dependencies
- Quantify re-entry, delays, errors and exceptions
- Surface business rules that have never been documented
What you receive
- Flow map
- Baseline measurement
- Constraint register
Validation gate
The flow is confirmed by the people who run it, and a business owner accepts responsibility for that shared reality.DECISION MADE
Prioritise
Choose an outcome, not a collection of features.
We isolate the first scope capable of producing a visible benefit without weakening the rest. Value, effort, risk and dependencies are made explicit so we can decide what enters the first cycle — and what waits.
What we do
- Compare expected impact with effort and risk
- Define an observable metric and its baseline
- Set the boundaries of the first flow to automate
- Decide where a human must remain in control
What you receive
- Target architecture
- Prioritised backlog
- Success criteria
Validation gate
The scope, inputs, outputs and definition of the expected result are understood without ambiguity.SYSTEM PROVEN
Build
Make the system visible before it is finished.
We work in usable, demonstrable increments. Business rules, integrations, error handling and traceability are developed together so reliability is never postponed until the end.
What we do
- Prototype the journey and critical decisions
- Connect systems in an isolated environment
- Log every action and handle failure cases
- Demonstrate the flow regularly to the people involved
What you receive
- Sandbox flow
- Decision log
- Test scenarios
Validation gate
Critical scenarios work with representative data, and every exception has a clear destination.CONTROLLED PRODUCTION
Deploy
Go live without taking a blind bet.
Going live is a controlled transition, not an improvised switch. We prepare access, acceptance testing, documentation, training and rollback before exposing the system to real volume.
What we do
- Limit and verify production access
- Run acceptance testing with business users
- Document operations, alerts and recovery
- Train owners and rehearse rollback
What you receive
- Deployment plan
- Operational runbook
- Access matrix
Validation gate
Acceptance has been signed off, owners are trained and the rollback procedure has been verified.DURABLE CAPACITY
Evolve
Keep the advantage after the first flow is delivered.
After launch, we observe real usage, system health and new constraints. The foundation can then support more volume, new flows and advances in AI without starting over.
What we do
- Track adoption, incidents and realised gains
- Maintain connectors, access and documentation
- Prioritise improvements from usage data
- Maintain an AI and technology watch relevant to the business
What you receive
- Health dashboard
- Evolution backlog
- Governance cadence
Validation gate
An owner, health indicators and a path for evolution are active and understood across the business.Before production
Fast to deliver. Solid to maintain.
Acceptance testing is not a formality at the end. Its criteria are defined during prioritisation, tested during construction and verified before launch.
PRODUCTION READINESS CONTROL 5/5 VALIDATED
- 01Business scenarios✓
Nominal journeys and critical exceptions are tested with realistic data.
- 02Security & access✓
Accounts are named, permissions are minimal and secrets remain outside the code.
- 03Observability✓
Errors are visible, qualified and routed to the right person.
- 04Documentation✓
The flow, its responsibilities and its operation can be taken over without hidden dependencies.
- 05Rollback✓
The response is known if production does not behave as expected.
RELEASE DECISIONReady for the real world.
GOWorking together
You keep the decision. We take responsibility for the system.
Business reality.
Your users explain the real work, validate the rules and decide priorities.
Design and execution.
We scope, build, test, document and secure the path to production.
Validation gates.
Every important decision remains visible, reasoned and connected to concrete evidence.
The right place to start
Bring the real problem. Let’s build the first proof.
A 30-minute conversation is enough to identify the first flow to observe and the people who should be involved.
Discuss my project