Current-state baseline
How the work actually runs today: volume, handling time, roles, delays, correction load, exceptions — and where the evidence is thin.
Loading — one moment.
For one bounded workflow, the Blueprint measures what the work costs now, what should change, what the controls must be, and what “working” will mean. Its job is a defensible decision — not a forced sale.
Modelled scenarioThe current first-cohort structure is deliberately narrow. The clock starts only when the prerequisites are real.
The output connects how the work actually runs, what it costs, what could go wrong and what is technically feasible. It is not a generic AI-opportunity report.
How the work actually runs today: volume, handling time, roles, delays, correction load, exceptions — and where the evidence is thin.
The trigger, the permitted inputs, the steps rules can own, the steps AI can help with, the human decisions, and where the work must stop.
What the workflow costs now, what is realistically addressable, the assumptions, conservative scenarios, and what recovered capacity would honestly be worth.
Data sensitivity, permissions, providers, regions, integration feasibility, client dependencies, and the questions that need a specialist before anything is built.
The representative test inputs, observable behaviours, tolerances, reviewers and evidence that will decide whether the installed system counts as working.
The reusable claims and findings your client work depends on: their sources, owners, permitted uses, current state, and what would invalidate each one.
Who should keep those claims current between engagements, derived from how fast your sources change, how often the work recurs, and whether a real internal owner exists.
Install, narrow the scope, use ordinary automation, configure a tool you already own, build it yourselves, defer — or do not proceed.
A responsible design phase protects you from paying for an elaborate answer when a smaller or existing one is better.
The workflow, the economics, the access and the control boundary support a separate installation proposal.
A smaller version reaches value sooner, or removes a dependency that would make the build fragile.
The job is predictable enough that AI would add cost and uncertainty without adding judgment support.
A platform you already pay for can solve the problem credibly, without a separate installed system.
Your team has the capability, and the Blueprint is enough to guide the internal build.
The evidence, economics, access, ownership or risk do not support implementation now.
The boundary keeps the decision honest and stops discovery from becoming an open-ended implementation engagement.
Representative workflow analysis, assumptions, dependencies, controls, evidence gaps and acceptance behaviours for the agreed job.
Integrations, migration, production configuration, new access, full security remediation and ongoing operation sit outside the Blueprint.
The no-fee Fit Call tests whether the job is important, bounded, owned and ready enough for a paid decision. It does not require confidential material.