The complete run-of-show and artifact reference for the AI Operating Sprint. This is the working document, not the public offer page.
The AI Operating Sprint
A five-day, fixed-scope workshop that transfers a working AI practice to an operating executive and their team. The first vertical is manufacturing, co-delivered with a lean/CIP partner. The core is vertical-agnostic.
Who it’s for
An operating executive (a COO, president, GM, or head of operations) who has personally discovered AI in the last few weeks, is getting real wins, and is about to hit a wall.
The wall, in one line
Everyone has a seat and nothing compounds. The company spread out.
The wall has four parts:
- The work depends on spreadsheet exports.
- The analysis lives on one desktop.
- The output is deep but unreadable to the leadership team.
- Conservative IT leaders and hungry executives have no trusted bridge between them.
What the team leaves with
- A live read-only connection to the relevant database.
- The operational KPI structure rendered in the review format the leadership team already uses (the bowler in manufacturing).
- A shared, governed setup with boundaries and owners documented.
- The practice itself, transferred, with a one-page maintenance rhythm.
This is practical delivery, not a course or an assessment. Anyone trained in this model of thinking can deliver the workshop.
Price
$9,000–$12,000, fixed scope, set during scoping.
The arc
Five working days over roughly eight calendar days, with soak days between. Each working day is a half-day clinic (education + hands-on, open forum, screens shared) and a half-day of office hours (personal projects, the AI lead on call). Up the left edge first (single-player, Days 1–3), then across the top (multiplayer, Day 4), then the read-back and handoff (Day 5).
Two projects are set up front with the stakeholder: a single shared project the group works on together (the multiplayer teaching vehicle) and a personal passion project each person picks (the single-player teaching vehicle).
Day 1 — the operation
Justin leads the review mapping; the AI lead runs the clinic; both observe and restructure live.
| Time | Segment | Purpose | Method | Output | Responsible |
|---|---|---|---|---|---|
| 0:00–0:30 | Set the two projects | Identify the shared project and each person’s passion project | Facilitated scoping with stakeholder | Two projects named, each person has a passion project | AI lead + stakeholder |
| 0:30–0:50 | The bowler, in your hand | Put the out-of-the-box product in the buyer’s hand Day 1 | Hand the bowler template against a dummy dataset | The product is concrete | AI lead |
| 0:50–1:50 | Walk the operation | Map the monthly review as it runs today; surface where AI analysis already lives, what depends on exports | COO shares screen, walks the review; Justin maps live on the board | Review map, the four-part wall in their language | Justin + COO |
| 1:50–2:20 | Individual reflection | Surface the shadow AI use in the team | Each person writes privately, then shares aloud | The shadow-use map | AI lead |
| 2:20–2:35 | Break | ||||
| 2:35–3:05 | Live demonstration: the AI lead’s reference setup | Make the top-left corner of the operating map visible once | Recorded demo + setup recipe (a substitute facilitator can build or be shown beforehand) | The operator’s practice, made concrete | AI lead |
| 3:05–3:35 | The operating map diagnostic | Place the company on the map; name “everyone has an account” as the structural failure; the wall becomes “wide, not deep” | Three dots on one picture; open forum | The diagnosis in their language | AI lead |
| 3:35–4:20 | Open forum: the gap | Break the chatbot frame in conversation | Group processes the gap between what they saw and what they do | The chatbot frame is broken | AI lead |
Afternoon — office hours. Each person starts their personal passion project. AI lead available. No group session.
Day 2 — the data and the rules
The AI lead leads the clinic; Justin translates into operational vocabulary.
| Time | Segment | Purpose | Method | Output | Responsible |
|---|---|---|---|---|---|
| 0:00–0:45 | Think first, by hand | The human does the thinking; the model does the generating | Each person takes paper: “what am I actually trying to decide from this review?” | Handwritten questions | AI lead |
| 0:45–1:15 | Restructure the question | Narrow, specific asks beat broad ones | Share what each wrote; AI lead restructures one live on screen | Restructured questions | AI lead |
| 1:15–1:45 | Craft a persona | The persona drops the model into the training data you need | Build a specific, named persona (cost accountant, ops reviewer, with blind spots) live on screen | A crafted persona | AI lead |
| 1:45–2:00 | Break | ||||
| 2:00–2:45 | Run the same question two ways | Feel the latent-space anchor by difference | Run the same question against the default assistant and against the persona; watch the output change | The difference, felt | AI lead |
| 2:45–4:00 | The data and the rules | Map systems, connections, data, constraints; make the conservative IT leader safe; name the four controls | IT leader shares screen; map live; name verified read-only, audit log, rollback path, 2 a.m. owner | The data map, the four controls named in the IT lead’s language | AI lead + IT leader |
Afternoon — office hours. Personal passion projects continue. Optional 1:1 slots for the conservative IT leader and the progressive exec, separately.
Soak day 1 — reflect and experiment
Participant time. The AI lead on call in office hours.
All day: reflect and experiment. The read-only bridge, where it is built, is a parallel activity with the IT lead.
Day 3 — the build, together
Co-delivered; the clinic gets intense. The Day 3 gate is the one objective test in the arc.
| Time | Segment | Purpose | Method | Output | Responsible |
|---|---|---|---|---|---|
| 0:00–1:15 | Personal projects, live | Apply the single-player skills (segment, sub-agents, judgment) to real work | Each person works their project, sharing screen in rotation; AI lead names what’s happening | One bounded step per person (in progress) | AI lead + participants |
| 1:15–1:30 | Break | ||||
| 1:30–2:30 | The shared project, live | Surface the multiplayer difficulty (naturally, not by lecture) | Group works the shared surface together | The difficulty, felt | AI lead + team |
| 2:30–3:15 | Restructure live: drift and restart | Teach the intuition (sense drift, redirect or kill and restart) | Stop when output goes wrong; share screen; name it; apply the restructure-live pattern | The intuition, taught in real time | AI lead |
| 3:15–3:45 | The judgment move, named | The model has no judgment; the human is the judgment agent; the 80/20 | Point to the moment a participant judged and name what they did | The judgment move, named | AI lead |
Afternoon — office hours. The gate is checked here. Each person shows the AI lead their finished bounded step. The AI lead checks the shared-surface difficulty is felt (observable: at least two collided, or one said “we’re stepping on each other,” or a merge/overwrite occurred). If the gate fails, adjust before Day 4.
Soak day 2 — reflect and experiment
All day: reflect and experiment. Hardening the shared setup and drafting the maintenance rhythm happen as parallel activities with whoever owns those surfaces.
Day 4 — shared and governed
The AI lead leads. The shared project becomes the multiplayer teaching vehicle.
| Time | Segment | Purpose | Method | Output | Responsible |
|---|---|---|---|---|---|
| 0:00–1:00 | Owned surfaces | Resolve the Day 3 difficulty by drawing boundaries | Name the difficulty; map who owns which surface live | Owned-surface map | AI lead + team |
| 1:00–1:30 | The LFW shape, named | One owner per surface; handoff via committed artifact with instructions; bounded persona per surface | Name the working shapes against their org | The pattern, named | AI lead |
| 1:30–1:45 | Break | ||||
| 1:45–3:15 | Governance negotiation, live | Place a human at every boundary; draft the boundary-and-owner map | Conservative IT leader + progressive exec in the room; share screens; draft live | The governance doc, drafted live (pedagogy) | AI lead + IT leader + progressive exec |
| 3:15–3:45 | The bottom-right trap, named again | “Everyone has an account” is where you go if you skip this; the shared substrate is the mayhem | Open forum; the group’s own near-misses on screen | The trap, named | AI lead |
After the live draft, the AI lead hardens the governance document (with the IT lead) into a signed, version-controlled control before Day 5.
Afternoon — office hours: the mystery. The group connects their agents to the bounded shared surface and works together to solve the mystery. If the mystery isn’t built, fall back to a second pre-shaped shared surface, or extend the morning’s governance work into office hours.
Day 5 — the read-through and the handoff
Co-delivered. 90 minutes with the full leadership team. No office hours.
| Time | Segment | Purpose | Method | Output | Responsible |
|---|---|---|---|---|---|
| 0:00–0:30 | The review running live | The practice, transferred | COO and team run it, not the AI lead; both projects shown | The practice, demonstrated | COO + team |
| 0:30–0:50 | The operating map read-back | Where you were, where you are, the next move | Three dots again; the path named | The read-back | AI lead |
| 0:50–1:15 | The runbook and ownership transfer | Prove the transfer; commit to maintaining or handing over | The maintenance rhythm (one page) + the runbook + the ownership-transfer sign-off | The artifacts signed | AI lead + IT leader |
| 1:15–1:30 | The handoff | It’s theirs; the honest recommendation | Facilitated; the retainer offered if honest read is that ongoing density would help | The decision | AI lead |
The artifacts
The bowler template
The tiered review format (tier one financials, tier two operational KPIs, tier three drivers), built against real data, in the format the leadership team already uses.
The governance map template
The boundary-and-owner map: who owns which surface, what crosses each boundary, which human makes the call at each one. Drafted live on Day 4, hardened into a signed, version-controlled control before Day 5.