How we work
We use AI-centric methods to move a priority use case from discovery and evidence to a decision and, when separately agreed, through implementation, capability transfer or ongoing support. The five services are buyer entry points, not a mandatory programme.
Four ways to engage an AI company
Strategic consulting, implementation support, sector-context solutions and ongoing work can each be contracted with different responsibilities. A focused diagnostic does not automatically include production delivery or managed operation.
AI-assisted research, scenario testing and evidence synthesis help frame value, architecture, governance and operating decisions. Human experts check sources and context; client leaders retain responsibility for consequential decisions.
Where separately contracted, this service translates a decision into a bounded prototype, integration, evaluation or control implementation. Technical owner, security and data boundaries, acceptance tests and handover are specified first.
This service adapts the AI method to the real sector workflow, affected people and operating constraints. These are prospective solution approaches, not claims of named deployments or regulator endorsement.
Only under a separate written agreement defining activities, responsible team, service hours, data handling, monitoring and incident boundaries, reporting, service levels and exit. No 24/7 platform service is implied.
The cross-service method
A service may begin at any relevant stage. We select and scope the stages the client actually needs; the route below is a decision discipline, not a fixed timetable, contract or automatic implementation pathway.
Begin with one priority AI use case, the intended outcome, the sponsor and the first blocked decision.
Decision gateIs this problem defined and material enough to scope?
Human ownerExecutive sponsor
Potential outputA likely starting point and the next decision. A conversation is not a diagnostic engagement.
Define the use, affected people, workflow, assumptions, measures and boundaries before choosing a solution.
Decision gateIs there a decision clear enough to assess?
Human ownerBusiness owner
Potential outputDecision brief or use-case charter, if contracted.
Examine value, consequences, data provenance, technology options, supplier evidence and operating dependencies relevant to that choice.
Decision gateWhat needs evidence, remediation or escalation?
Human ownerBusiness owner, with relevant data, risk or technology owner
Potential outputScoped assessment, dependency map or evidence request.
Design decision rights, controls, technology criteria, data actions, human–AI work or an agreed non-production evaluation where it helps reduce uncertainty.
Decision gateIs the design owned and proportionate enough for a controlled delivery decision?
Human ownerDesignated approval or control owner
Potential outputBlueprint, control design, options record or a separately scoped prototype plan.
Decide whether to approve, revise, hold or stop. Where separately contracted, ISOVIA can support AI implementation, integration or testing under explicit technical responsibilities.
Decision gateAre conditions, access, acceptance tests and escalation agreed?
Human ownerClient accountable executive and contracted delivery lead
Potential outputDecision paper and, only where contracted, technical delivery artefacts.
Hand over rationale, roles, review routines, challenge routes and an improvement backlog to the team that will own the work.
Decision gateHas the receiving team accepted ownership?
Human ownerReceiving service owner
Potential outputHandover and enablement plan.
Review changed decisions or, where a separate managed scope is agreed and resourced, operate clearly defined AI service activities with stated service levels and an exit route.
Decision gateAre the purpose, team, hours, controls, reporting and exit explicit?
Human ownerClient service owner and contracted ISOVIA lead
Potential outputA separately agreed advisory or managed-service record. Continuous 24/7 coverage is not included by default.
Choose the first door
You do not need to buy all five. Start where the first blocked decision lives; adjacent services become relevant only when the evidence points to them.
Which opportunities deserve funding now, and what has to be made ready first?
Explore the service 02 / 05Who can approve this use case, under which conditions, and with what evidence?
Explore the service 03 / 05Should you build, buy or combine - and what must remain portable?
Explore the service 04 / 05Can your team trust, locate, explain and correct the data this use case needs?
Explore the service 05 / 05Who makes the final call, who challenges the system and who improves it after launch?
Explore the serviceCommercial stages
A conversation clarifies the need. Any diagnostic, blueprint, implementation or ongoing activity needs a statement of work with deliverables, owners, acceptance criteria, access, terms and timing agreed case by case.
01
Clarify the use case, sponsor, intended outcome and blocker. This is not an assessment delivered for free.
02
Agree a defined evidence review and the decision it supports. The findings may support proceeding, fixing the gaps, deferring the work or stopping it.
03
Design selected strategy, governance, data, technology or operating artefacts with named owners and an acceptance route.
04
Only when separately agreed: bounded technical work with data and security obligations, human review, tests, escalation and handover.
05
Explain the reasoning, prepare the people taking over the work and record what needs to be reviewed or improved next.
06
Only with a defined service boundary, coverage hours, responsibilities, controls, reporting, service levels and exit in a written agreement.
Aegis governance lens
Aegis links a use case to its owner, risk, policy, controls, human oversight and evidence when useful. It informs how we ask governance questions. It does not itself promise deployed software, automatic scoring or continuous legal compliance.
Explore the Aegis approachFounders’ wider thinking
Nandakumar and Archana’s forthcoming Kinetic AI Transformation Framework book examines five enterprise-architecture layers. They are a way to reason across systems and accountability, not a substitute for our five scoped services, the Aegis governance approach or a contract for delivery.
Read the public framework previewBring one AI use case, an accountable sponsor and the first blocker. We will establish whether a focused, separately scoped next step is useful.