Working session
Sit in on the work we'd do together
These are the three conversations I have with every leadership team: which work should AI take, how far can it go on its own, and what the first eight weeks look like. Pick one and try it with your own numbers. It runs in your browser, with no live AI and no real systems.
Illustrative model · Runs in your browser · No production actions
| Work | Volume/mo | Min/item | Disposition | Human h |
|---|---|---|---|---|
Classify and route incoming requests Accountable: Service operations lead · low consequence | 400 h → 400 h | |||
Answer order and account status questions Accountable: Service operations lead · low consequence | 350 h → 350 h | |||
Research complex technical issues Accountable: Tier-2 support manager · medium consequence · judgment | 525 h → 525 h | |||
Approve goodwill credits and refunds Accountable: Customer finance owner · high consequence · judgment | 67 h → 67 h | |||
Handle at-risk customer escalations Accountable: Customer success director · high consequence · judgment | 120 h → 120 h | |||
Compile weekly status report by hand Accountable: Service operations lead · low consequence | 16 h → 16 h |
Scroll the table sideways to see every column.
Today: 1,478 h human time
Designed: 1,478 h human time (bar scaled to today)
- Eliminate
- Simplify
- Retain human
- Automate
- Augment
- Agentify
- Autonomize
- Capacity returned
- 0 h
- Full monthly cost
- $83,710 → $83,710
- Cost per outcome
- $13.95 → $13.95
- Authority
- No gaps flagged
Assumptions & formulas
- Human hours = volume × minutes × time factor × human share + reviewed items × review time + exceptions × exception time.
- Full cost = human hours × labor rate + model/tool cost + control cost + rework on failed outcomes. Cost per outcome = full cost ÷ resolved issues.
- Machine work is always sampled for review (minimum 2%); high-consequence Agentify/Autonomize work is reviewed in full.
- Capacity returned is time, not guaranteed cash savings. It only becomes value when redeployed or when the cost base changes.
| Disposition | Human share | Review | Tool × | Control $/mo | Success |
|---|---|---|---|---|---|
| Eliminate | 0 | 0 | 0 | 0 | 1 |
| Simplify | 1 | 0 | 0 | 0 | 0.97 |
| Retain human | 1 | 0 | 0 | 0 | 0.97 |
| Automate | 0.05 | 0.05 | 0.3 | 400 | 0.95 |
| Augment | 0.6 | 0 | 0.6 | 300 | 0.98 |
| Agentify | 0.15 | 1 | 1.2 | 1500 | 0.93 |
| Autonomize | 0.02 | 0.25 | 1.5 | 2500 | 0.9 |
The seven dispositions are choices made per piece of work on evidence and consequence, not a maturity ladder. From the book
Track record
Innovations & invention
FinOps by Design
Conceived the concept and led the team to second place among 266 entries in the Atos and Microsoft AI Hackathon, securing R&D support.
Pursuit Factory
A repeatable approach for shaping and running major enterprise pursuits, connecting solution leaders, architects and delivery around a clear client outcome.
Growth Engineer & forward-deployed consulting
Built practice approaches that embed solution leaders with clients to move from problem to working prototype quickly.
Agentic AI prototypes
Rapid concepts for intelligent workflows, designed around practical adoption, human oversight and delivery feasibility.
Patent · Named co-inventor
System and method for allocating resources and managing a cloud based computer system
International application PCT/US2014/072509 · Publication WO2015103113A1 (2015) · Stratus Technologies
View publication ↗