Ten sales AI implementations on two pages, ordered by how many registers each one has to read.
Each entry carries the row it comes from in the stack's 100-activity map, the mode it runs in, the difficulty, a deploy estimate, the impact and the risk posture it stops at. Seven run agent-led with review and three are AI-assisted, and the card shows that split rather than smoothing it: three of the ten produce an input to a person's work, and the other seven produce a verdict somebody then acts on. A verdict is what earns a review step.
Three of the ten in full, and all ten named.
All ten sit on rows the map rates 5 out of 5 for impact, so the rating separates nothing here. The order does, and it turns on registers rather than on ambition. The three below are the two builds week one is made of, which read no register at all, and one of the two that read all four and whose output a customer can hold you to.
Pipeline Hygiene AuditorAgent-led + review
Close dates in the past, next steps with no date, amounts with no currency and owners who left, counted and named by owner.
Map row 32 · Medium · about 3 hours · impact 5/5 · it flags and never corrects, because a close date moved by a bot deletes the slip
Weighted Forecast AssemblerAgent-led + review
The roll-up as the record stands, arithmetic shown, every category total reconciled to the sum of its deals.
Map row 72 · Medium · about 3 hours · impact 5/5 · it feeds a number a person signs; calling that number to the business is a separate row and does not ship
Proposal Draft AssemblerAI-assisted
Scope, pricing, terms and references each from an approved source, and a visible blank where there is none.
Map row 48 · Medium · about half a day · impact 5/5 · a blank is a question a human answers; an invented figure is a commitment nobody noticed making
The other seven, by name
Opportunity Qualification Challenger · Account Research Dossier Builder · Meeting Preparation Pack Builder · Discount and Margin Guard · Renewal and Expansion Signal Analyst · RFP Response Compliance Checker · Pipeline Review Brief Builder
On the card each of those carries the same six fields as the three above: the map reference, the mode, the difficulty, the deploy estimate, the impact and the risk posture it stops at.
What the deploy estimates assume
These figures are estimates from the same build pattern on comparable agents, timed while building the Finance and IT stacks. Estimated build time for one experienced sales operations analyst following the stack's build sheet, excluding organizational approvals and excluding your own prep: extract shapes, stage exit criteria, the margin floor and its cost base, the four registers. The prep is usually the longer half, and the card says so. Time your first build and replace the estimate with your own number, which is the one your next nine builds should be planned against.
The three flagship builds stop exactly one step short of the commitment.
Issuing the proposal. Sending the quote. Approving a discount above the delegated threshold. Agreeing a non-standard term. Promising a delivery date or a service level. Calling the forecast number to the business. Bulk-correcting CRM fields after an audit. Asserting a certification or a regulatory capability. Naming a reference customer. Stating a roadmap item as a present capability.
Those are not late-roadmap items waiting for a better model. Five of them are the second half of a split the 100-activity map makes on purpose, and each of the three implementations above stops exactly one step short of the commitment. That gap is the product rather than a limitation of it.
Then notice what is missing from the four registers, which the card states plainly: a delivery date has no authority source at all, which is why an agent can repeat a date back to you perfectly and still be wrong to commit it. Nothing here prepares a safety authorization either. No output qualifies a product, a service or a person for a safety-critical or regulated duty, and a certification invented to win an RFP is the sales version of signing a permit nobody inspected.
commitments a customer can hold you to, and an agent states none of them on its own. A price. A term. A delivery date. A statement of capability. A named customer reference. Four of the five resolve to one of the four registers. The delivery date resolves to none of them, so an agent can repeat a date back to you perfectly and still be wrong to commit it.
Ordered so week one does not wait on a data project.
The sequence turns on one property: how many of the four sourced-assertion registers a run has to read. They are the price book, the standard terms, the capability register and the reference register, and two of the four are the ones you are least likely to already hold in a form an agent can be handed. So the builds that need them go last, and the weeks before them are built to be worth the effort on their own.
Week 1, the internal weekTwo builds
The hygiene audit tells you which fields the forecast is about to be computed off, then the assembler computes it with the arithmetic shown. Neither one reads a register, so the first week does not wait on a data project.
Week 2, the commercial edgeThree builds
A margin test that states the breach instead of looking for a construction that justifies it, a review brief that cuts the meeting to three deals, and a qualification challenge that names what is missing rather than scoring it. Still no register.
Week 3, the customer-facing weekThree builds
The dossier, the renewal signal and the meeting pack. Two of these read one register each, which is where you find out what your capability and reference entries actually cost to assemble, on a small surface.
Week 4, the commitment weekTwo builds
The proposal assembler and the RFP compliance checker, the two that read all four sources. These are the only two whose output a customer can hold you to, which is why a buyer builds them last. The card names which build goes in which week.
Two of the four registers are the ones you are least likely to already hold in a form an agent can be handed.
The card names those two: the capability register and the reference register. They live in people's heads and last year's slides, and building them is the real cost of month one.
It is on the free card and it is on the paid page for the same reason: a buyer who discovers it in week four is right to be annoyed. Weeks one and two are sequenced to be worth the effort before that bill arrives, which is also why the two builds that read all four sources sit at the end rather than in the demo.
registers you are least likely to already hold in a form an agent can be handed. The work is writing down what you actually deliver and who has actually agreed to be named, with an owner and an effective date on every entry.
Free. Pay $0 at checkout and the PDF lands in your library.
Licensed, not sold. Free resources ride the same License & Terms as everything else here (Article 3.1); the full notice is at the foot of this page.
Practitioner guidance, not legal, commercial or contractual advice. AI prepares the seller. The seller decides.
Two pages, PDF. Pay what you want if you like it, starting at $0.
The card is the sequence. The Stack is the build.
The card is the two-page table, the week-by-week order, the month-one register cost and the list of what the ten stop short of. It is a sequence, not a build. The Stack is where all ten arrive as paste-ready builds with their read scope and their refusals written down, and where the fixtures and the answer keys live so you can grade your own runs by hand.