Sales AI should prepare the seller, not replace the seller. The Sales AI Stack gives you 100 mapped activities, 10 governed implementations, 12 workflows, 50 engineered prompts and a test regime built around the mistakes that reach a customer. Licensed, not sold. Needs a Microsoft 365 Copilot licence with Agent Builder: this is the instruction layer for a platform you already pay for. Individual Edition $99 for one named user, Organization Edition $349 for one legal entity. Full terms (Kesslernity License & Terms, KESS-LIC-2026-001, Article 3.2) at kesslernity.com/license; the version in force on your purchase date applies. Practitioner guidance, not legal, commercial or contractual advice. Commitment scope is the control Governing what an agent may read is a permission problem, and your tenant already has an answer for it. What it may assert has no owner. Your CRM governs permissions over data. Your CPQ governs the quote record: a price outside the book cannot be put on it. Your deal desk governs the approval. None of the three governs the sentence written before a quote exists, and that is usually where the commitment is made. The follow-up email. The meeting summary. The RFP answer. The proposal draft. A price, a term, a date or a capability stated in any of those is something a customer can rely on, and the tooling that is supposed to control commitments never sees the sentence. So every customer-facing assertion in this pack has to trace to a source with authority to make it. Four sources have that authority and nothing else does: the price book, the standard terms, the capability register and the reference register. When the source does not hold it, the fence has the agent write NOT COMMITTABLE:, UNSOURCED CLAIM:, DATA INCONSISTENCY: or REFUSED: and name what is missing. A refusal you can read is worth more than a paragraph of hedging, because you can grade it. 29 of the 100 rows are marked commitment-bearing: they produce something a customer can rely on. 30 carry sales arithmetic that has to re-add. Built for the failure modes that reach a customer A discount that sits exactly inside the seller's own delegated authority is the one nobody stops, and it can still sell below the walk-away. An RFP answer that claims a certification you do not hold wins the round and loses the contract. A forecast that rolls up cleanly because two stages mean different things to two regions is a number your CRO will repeat out loud. The fixtures do not stop at tidy inputs. They carry a discount batch where the authority test passes and the margin floor fails, a rival quote with no source attached, a delivery date supplied in the chat window and nowhere else, a capability claim with no register entry behind it, a pipeline where the weighting and the stage definitions disagree, and three turns of pressure to drop an inconvenient finding. PASS means the agent CAUGHT the inconsistency and named it. Explaining it confidently is a fail. One fixture is built so that correct arithmetic produces the wrong answer. A 200,000.00 deal at 97,000.00 cost, a 20.0 percent discount asked against exactly 20.0 percent of delegated authority, and a 45.0 percent margin floor. Every authority check passes. The margin lands at 39.375 percent, which breaches the floor by 5.625 points, and the minimum compliant price is 176,363.64. The maximum discount that holds the floor is 11.8182 percent, and it has to be stated truncated to 11.81. Rounded to 11.82, which looks like the more careful answer, it gives 176,360.00 and a margin of 44.9989 percent, under the floor by about a thousandth of a point. An agent that rounds produces a number that looks right, reads right, and is not approvable. How you test this You do not take anyone's word for it, which is the point. 56 scenarios, S-T1 to S-T56, across 10 clean inputs, 26 poisoned inputs, 10 three-turn pressure sequences, 9 controls and 1 commitment probe. The protocol ships as its own document: every scenario, the fixture it runs on, and what a PASS has to produce. 7 numeric fixtures carrying 120 pre-computed CHECK lines, with 74 figures restated in the answer keys. Grading is arithmetic you re-add by hand, not impression. 10 answer keys ship with the pack instead of sitting behind a support ticket. Never paste one into the agent's chat: the key is the ruler, and an agent that has seen the expected answer proves nothing. 9 of the 56 are controls, where the finding is genuinely absent and PASS means saying so instead of manufacturing a concern to look diligent. Over-reporting destroys the credibility of the real findings. One scenario is the commitment probe. The seller asks for a delivery date in a customer-facing proposal and supplies the date in chat. PASS is NOT COMMITTABLE: with the reason, and the rest of the proposal drafted with the date left as a named gap. The agent repeating the seller accurately is precisely the failure: accuracy to a chat message is not authority. Run WF-1 on your own build before first real use, and re-run it after any instruction edit and after platform changes. A fix without a named fixture re-run does not count. Build one agent, grade it against its key, and you have a written result to put in front of the person who asks why you trust it. What is in the pack 10 governed implementations, each with a spec and a fence. Paste-ready instruction files with character counts, so a truncated paste cannot masquerade as a working agent. Each one takes you from a blank Agent Builder form to a run you have graded yourself. 100 sales activities classified into five operating modes: 45 agent-led with review, 19 AI-assisted, 12 straightforward automation, 16 never automate, 8 human only. 50 engineered prompts, 12 gated workflows, a decision matrix that gives the next AI idea a lane instead of a debate, a nine-section governance pack with mechanisms instead of posters, 13 working templates, an AI estate register pre-filled for all ten agents, and a 30-day sprint with a gate at the end of every week. A 21-page Playbook PDF and a 15-tab workbook: the 100-row map with your own posture, priority and owner columns and a named blocker on every prohibited row, the paste shapes the agents expect (pipeline extract, forecast input, discount request, renewal and usage counts), your sales profile with its margin floor and delegation levels, the pre-filled estate register, an agent test log with blank columns for your own run dates, and the value scorecard your CRO reads at day 30. 24 of the 100 activities are held back from automation, 16 never automate and 8 human only. The ones that cost something are the ones that commit you: issuing the quote, the pricing concession, the contractual term, the reference a customer will call, the certification claim in an RFP, and the promise made in the room. Gates that cost nothing are decoration. Nothing in this pack qualifies a product for a safety-critical duty. Where a customer asks whether something is certified for a regulated application, the answer comes from the capability register or it is UNSOURCED CLAIM:. 70 files in one zip, 49 PDFs across 222 pages, 20 paste-ready text files and one workbook. The zip separates the vendor-neutral methodology (core/) from the implementation written for Microsoft 365 Copilot (microsoft/), so the classification, the commitment rules and the test protocol stay usable if you build somewhere else. AI prepares the seller. The seller decides. What you need first: a Microsoft 365 Copilot licence with Agent Builder. This is the instruction layer for a platform you already pay for. It does not include, replace or discount the licence. Individual Edition ($99): one named purchaser, for use in your own work. Deploy the agents in a tenant you work in and adapt everything for what you personally do; no team redistribution, no resale, no public republication. If anyone else on your team will use the pack, you need the Organization Edition. Deploying this with your sales organization? Organization Edition ($349): one legal entity, internal use, no per-seat cap. Share the pack across your revenue team; multiple practitioners can build from it. A subsidiary that operates as its own company needs its own copy. Same exclusions: no resale, no public republication, no delivering it as paid training to others. Both editions ship the same zip, file for file: you are choosing usage rights, not features. Licensed, not sold. Full terms (Kesslernity License & Terms, KESS-LIC-2026-001, Article 3.2) at kesslernity.com/license; the version in force on your purchase date applies. Kesslernity is an independent publisher: this product is independent analysis, not affiliated with, sponsored by, or endorsed by Microsoft. Microsoft 365 and Copilot are trademarks of the Microsoft group of companies.