Enterprise architect
Validated assessment
The 3 pilot candidates above are scored from model structure alone. A validated assessment walks each one with its process owner, checks the task classifications against the real exception mix and volumes, and turns directional scores into a sequenced roadmap with owners and dates — including the 3 processes parked at “Assessment needed”.
First movePut this report and estate.json in front of the process owners behind the Pilot candidate tier; the assessment starts from their corrections.
Platform engineer
Engine parity pilot
Run Card Dispute Intake & Resolution on Handeware Engine as-is — the same BPMN, the same DMN tables, your handlers behind each service task. The engine is event-sourced and replayable with suspend / approve / resume at every human gate, so parity against the incumbent runtime is read from the evidence bundle, not from a demo.
First movehandeware-run start card_dispute_intake.bpmn --handlers handlers.py --dmn-dir ./dmn then handeware-run audit <run_dir> and compare the bundle with the incumbent's case log.
Risk & controls
Governance pack
4 touchpoints across the estate are classified agent-with-HITL and none are human-only. The governance pack maps every agentable touchpoint to a named approval gate, an event-sourced evidence bundle and a replayable decision log, so the first agent goes live inside existing model-risk and change controls rather than beside them.
First moveWalk the agent-with-HITL rows in the process cards below the fold; each one needs an approver role and an evidence requirement before a pilot starts.
All three steps run on the same artifacts this scan already produced — no new data leaves the sandbox.