The current workflow is implicit
Rules, workarounds, exceptions, and ownership live in individual knowledge, making automation brittle or incomplete.
03 · AI capability
Intelligent Automation
Combine software, workflow rules, integrations, and AI where they can reduce repeated effort without hiding ownership or exceptions.
The context
Begin with the work, evidence, and consequence.
Intelligent automation works best when the operation is understood end to end. The design should separate stable rules from probabilistic interpretation, keep material decisions reviewable, and make exceptions visible to the people responsible for resolving them.
AI is used for deterministic work
A probabilistic model adds cost and uncertainty where ordinary validation, rules, or integration would be clearer.
Exceptions disappear
The normal path moves faster, but low-confidence results, provider failures, invalid data, and manual intervention have no designed destination.
Areas of attention
Connect AI behaviour to a dependable product and operation.
The appropriate techniques, controls, and delivery depth follow the use case. These areas keep the technical work connected to the people and decisions around it.
01
Workflow and decision mapping
Trace triggers, roles, systems, data, rules, judgement, controls, hand-offs, and exception paths.
02
Automation pattern selection
Choose between application logic, integration, rules, extraction, classification, generation, and human review for each step.
03
Human-in-the-loop design
Provide confidence-aware review, understandable evidence, correction, escalation, and accountability where judgement remains essential.
04
Operational control
Design queues, status, audit records, monitoring, retries, reconciliation, and safe recovery around the automated flow.
Potential outputs
Create evidence and artefacts the team can use.
Outputs are shaped around the decision and engagement stage. Each should have a clear audience, purpose, owner, review criteria, and stated limitation.
- 01Current and target workflow map
- 02Automation-candidate assessment and prioritisation
- 03Rules, AI, integration, and human responsibility design
- 04Working automation increment with exception handling
- 05Monitoring, audit, recovery, and operational ownership guidance
Evaluation and responsibility
Make inputs, useful evidence, and boundaries explicit.
AI quality cannot be separated from its context of use. Representative inputs, subject expertise, realistic scenarios, and accountable owners are part of the engineering work.
What we need to understand
- People who perform and manage the current workflow
- Representative documents, messages, cases, rules, and exceptions
- Systems, APIs, access, data definitions, and provider constraints
- Quality thresholds, control needs, service expectations, and escalation responsibilities
Evidence of a useful direction
- Automated steps are justified against simpler alternatives
- Important exceptions have an explicit owner and route
- Low-confidence or high-consequence results receive appropriate review
- Operators can inspect, correct, reconcile, and recover the workflow
Important boundary
Automation cannot repair unresolved policy, inconsistent source data, or unclear process ownership on its own. Some variable or consequential decisions should remain with an informed person.
Working path
Reduce uncertainty before expanding commitment.
The exact sequence depends on the use case and current evidence. Each stage should leave a reviewable result and an informed choice about the next step.
Observe
Map the real workflow, including workarounds, variations, controls, and failure recovery.
Separate
Distinguish deterministic rules, integration needs, model-assisted interpretation, and human judgement.
Implement
Build a bounded flow with validation, review, status, audit, and representative exception scenarios.
Operate
Measure workflow behaviour, review corrections and failures, and evolve the automation under clear ownership.
AI capability questions
Important details to clarify.
Early questions should expose the task, evidence, uncertainty, and responsibilities that shape a safe and useful implementation.
Is intelligent automation the same as robotic process automation?+
It can include RPA, but may also use APIs, application logic, workflow engines, document processing, machine learning, and generative AI. The appropriate mechanism follows the process.
Can AI make decisions automatically?+
It can support or automate some bounded decisions, but the consequence of error, available evidence, explainability, policy, regulation, and appeal or review needs determine the appropriate level of autonomy.
What happens when confidence is low?+
The workflow should define a review, clarification, fallback, or rejection path rather than forcing every input through the normal automated route.
Can we automate a process that is still changing?+
A changing process can be supported, but tightly automating unstable rules may create rework. It is often better to improve visibility and structure before adding deeper automation.
Start a conversation
Let's discuss intelligent automation.
Share the workflow, evidence, people affected, and the outcome you want. We can help frame a responsible first question and bounded next step.