An AI sales playbook maps specific prompts to every stage of the deal cycle — prospecting, discovery, evaluation, proposal, negotiation, and close — so AI works systematically instead of reactively. Most reps reach for AI only when stuck on a draft; the reps building the biggest pipelines run it at every stage, and the compounding is where the advantage lives.
Stage 1: Prospecting
The prospecting job is coverage with quality: identifying accounts worth pursuing and reaching them with something specific. Use AI for ICP fit verification, account research briefs, personalized first touches, and follow-up sequences.
Prompt: "You are a sales research analyst. Assess [COMPANY] against this ICP: [ICP CRITERIA]. Return a fit score with reasoning, the most likely business pain given their industry and stage, one recent trigger event, and a one-line personalized angle for a first touch."
The output turns a raw account list into a ranked queue with an angle attached to each name — the difference between fifty generic sends and twenty targeted ones.
Stage 2: Discovery
Discovery prep and debrief are the two highest-payoff AI moments in the entire cycle. Before the call, build a hypothesis: likely pain, financial impact, why now, and the first question that tests it. After the call, run the transcript through structured extraction — MEDDPICC scorecard, buying signals, risk flags, follow-up draft. The call itself stays fully human; AI owns the preparation and the paperwork on either side of it.
Stage 3: Evaluation and multi-threading
Mid-funnel is where deals quietly die: a single-threaded champion, an unengaged economic buyer, a technical evaluator with an objection nobody surfaced. AI helps you map and work the committee.
Prompt: "Based on these call notes and this org context: [PASTE], build a stakeholder map for this deal. For each known stakeholder: their likely priority, their probable objection, and one specific way to engage them. Then list the roles typically involved in a [DEAL TYPE] decision that I have not yet identified."
Run this monthly per active deal. The unidentified-roles list is usually the most valuable output — it names the person who will appear in week nine and stall everything.
Stage 4: Proposal and business case
The proposal's real audience is the meeting you are not in. Write it so your champion can present it without you: an executive summary in the prospect's own language, the problem stated with their numbers, and a one-page version for forwarding. AI drafts fast here, but feed it verbatim prospect language from your discovery extractions — that is what makes the document sound internal rather than vendor-written.
Stage 5: Negotiation and close
Late-stage work rewards preparation over improvisation. Build a negotiation brief before the call: your walk-away, their likely asks, trade options ranked by cost to you and value to them. Pair it with a mutual action plan that puts dates on every remaining step — legal, security, signature. Most slipped closes trace to an unowned step nobody wrote down, and the action plan is the fix.
Stage 6: Post-close handoff and expansion setup
The playbook does not end at signature. The two weeks after close set up either a clean expansion path or a renewal fight eighteen months later. Use AI to draft the handoff brief for your customer success counterpart — the promises made, the metrics the buyer committed to, the stakeholders and their individual wins, the risks flagged during evaluation. A structured handoff means the CS team starts with deal reality instead of reconstructing it, and it means the expansion conversation a year from now opens with documented outcomes rather than a cold re-discovery. Five minutes of extraction here protects months of relationship equity, and it is the cheapest pipeline you will generate all year.
Making the playbook stick
The stage-by-stage approach works when it becomes routine rather than rescue. Attach each prompt to a trigger you already have: new account in the queue means fit assessment; calendar invite means hypothesis brief; call ended means extraction; stage change means stakeholder map refresh. Reps who wire prompts to triggers stop deciding whether to use AI — and the consistency, not any single output, is what compounds into pipeline.
Frequently Asked Questions
Which stage should I start with if I'm new to AI in my sales workflow?
Discovery debrief. It has the fastest payoff, the clearest before-and-after, and its outputs — verbatim pain language and MEDDPICC gaps — feed every later stage of the playbook.
How much time does the full playbook add per deal?
It subtracts time. Each stage's prompt replaces a slower manual task — research, note-writing, proposal drafting — and reps typically report saving several hours per week once the triggers become habit.
Do I need different prompts for different deal sizes?
The structure holds at any size; the depth scales. SMB deals might use the prospecting and discovery prompts only, while enterprise cycles justify the full stakeholder, business-case, and negotiation stack.
Put It to Work
Every stage of this playbook maps to a collection in the Promptifi library — 2,900+ prompts across 16 categories, organized by exactly these deal moments. Browse the library and wire the playbook into your active deals.