Cycle time
How long does the workflow take from trigger to accepted outcome?
Case resolution · audit prep · incident recovery · approval routingValue engineering
Build the business case from completed workflows and accepted outputs.
Use completed workflows, accepted outputs, operating cost, adoption, and exceptions to test whether the proposed automation is worth running.
Measurement model
Instrument the workflow before you automate it.
Establish a baseline, define the accepted outcome, then compare human-only, assisted, and governed-agent performance using the same unit of work.
Cost to complete
What labor, model, tool, and rework cost is consumed per completed workflow?
Loaded labor · token spend · vendor overlap · escalation effortQuality and trust
How often is the first output accepted, cited, policy-compliant, and complete?
Acceptance rate · citation coverage · exceptions · reopen rateAdoption and reach
Are people repeatedly using the workflow, and is value spreading across teams?
Weekly active users · repeat runs · teams live · workflows reusedInspect the assumptions
A model your CFO can challenge.
Keep adoption, quality, implementation effort, inference cost, and change-management cost visible. The model should become more conservative as uncertainty rises.
Annual value(workflow volume × net time saved × loaded cost × adoption)+avoided rework and risk cost−platform, model, and operating cost
Illustrative worked case
Show the arithmetic, then challenge every input.
This fictional service-operations example estimates gross capacity value. It is not a customer result, savings guarantee, or ROI claim. Implementation, platform, model, oversight, and change costs still need to be entered.
Count accepted outcomes, not agent starts
Subtract review, correction, and escalation time
Use Finance-approved fully loaded cost
Use repeat production usage, not pilot access
Illustrative gross annual capacity value$120,900
10,000 × 18/60 × $62 × 65%. Deduct full operating cost and validate whether released capacity converts into lower spend, more throughput, faster revenue, or better service.
Download the assumption modelPilot scorecard
A disciplined path from baseline to scale decision.
Baseline
Observe volume, cycle time, handoffs, errors, and policy exceptions.
Shadow
Run the agent without writes and compare its recommendation to accepted work.
Assisted
Let users approve outputs and measure acceptance, edits, and time saved.
Bounded action
Enable low-risk actions behind clear policies and approval thresholds.
Scale review
Expand only when value, trust, adoption, and operating cost meet the charter.
Architecture session
Bring one consequential workflow. Leave with a governed agent blueprint.
We will map the context, systems, decisions, controls, actions, and success measures together.