A clear engagement from an operational question to a usable decision
Codance keeps each launch engagement proportionate by defining one workflow, one principal business function and one decision the work must support. Where a wider question spans several workflows, the relationships can be considered and a staged or separately scoped route agreed.
1. Begin with the decision
Tell us what operational area concerns you and what decision you need to make. Codance can help establish the most useful workflow boundary.
Useful starting questions include:
- Should we invest in AI for this process?
- Which part should we improve first?
- Are we ready to commission a detailed design?
- How should AI, software and people work together?
- Is a supplier proposal addressing the real operational problem?
Codance confirms whether the Readiness Assessment, Integration Blueprint or a separately scoped engagement is the right fit.
2. Define the workflow boundary
Together we identify:
- the event that starts the workflow;
- the useful outcome it should produce;
- the event that proves completion;
- the people, information and systems involved;
- the business function and exclusions;
- the person authorised to commission and describe the work.
“Our whole business” is not one workflow. A clear boundary produces a more useful answer.
If several workflows are connected, Codance can record their dependencies and recommend a staged scope without pretending that they can all be analysed within one launch package.
3. Complete the structured questionnaire
The questionnaire gathers the operational evidence needed for meaningful analysis. It asks what really happens, where information comes from, how decisions are made, what exceptions occur and what improvement would matter.
It is not a test. Honest uncertainty, estimates, unofficial routes and “it depends” are useful answers when they reflect reality.
4. Supply a representative evidence set
Depending on the service, Codance reviews up to five or ten relevant artefacts. These may include anonymised examples, forms, screenshots, checklists, process notes, system descriptions, measurements, diagrams or supplier proposals.
We do not need a large data dump. We need enough representative evidence to understand the normal path, important exceptions and limitations.
Do not provide passwords, keys, tokens, unnecessary personal information or material you are not authorised to share.
5. Discovery and clarification
Codance reviews the questionnaire and evidence, then uses the included discussion to resolve questions that could materially change the conclusion or design.
The work focuses on:
- the real current workflow;
- information and operational truth;
- decisions, authority and exceptions;
- delay, repetition, rework and hidden labour;
- the role of AI, deterministic rules and human judgement;
- risk, failure, fallback and recovery;
- value, measurement and acceptance.
6. Structured analysis
Codance distinguishes four things that are often blurred together:
Process improvement
Changes to ownership, sequence, information or policy that should be resolved before technology is added.
Deterministic automation
Work that follows dependable rules and should be handled predictably by conventional software.
AI assistance
Classification, extraction, retrieval, comparison, summarisation or drafting where probabilistic assistance creates useful value and uncertainty can be controlled.
Human authority
Consequential judgement, approval, exception resolution and commitments that remain the responsibility of named people.
7. Internal quality review
Before delivery, Codance checks that:
- findings are supported by evidence or labelled as assumptions;
- the output answers the agreed decision;
- limitations and confidence are explicit;
- AI has not been recommended where it adds no value;
- human authority, audit and fallback are preserved;
- the client can act on the result;
- the document is coherent, readable and correctly rendered.
8. Delivery and clarification
You receive the agreed AI Workflow Readiness Report or AI Integration Blueprint through the authorised channel, together with the principal finding and recommended next action.
Each launch package includes one revision for factual correction or agreed clarification and one post-delivery clarification response. Substantive new scope is agreed separately rather than silently absorbed.
What happens next?
The answer depends on the evidence. You may:
- improve the workflow or information first;
- proceed to an Integration Blueprint;
- brief developers or suppliers;
- run a shadow-mode evaluation;
- begin a human-approved pilot;
- use deterministic automation instead;
- decide that AI is not currently justified.
The engagement succeeds when it gives you a clearer, safer and commercially useful decision—not when it creates a predetermined technology sale.
Bring one real operational question
If you can describe where work is going wrong, what outcome should improve and why the decision matters, you have enough to start the conversation. Codance can help establish the right workflow and engagement boundary.