Turn a Buyer AI Agent Threat Assessment into a Sales Playbook Rewrite

The task

Account executives and sales enablement leads need to update their discovery questions and value-prop talking points for a world where buyers increasingly send an AI agent to do first-round research. This is the "if our prospect's AI shortlists us before a human ever reads our site, does our playbook still work?" exercise — and it's one you can run in an afternoon instead of booking a quarterly offsite.

Before AI

A playbook refresh usually means pulling the current discovery guide, interviewing three to five top reps, mapping their questions to the latest buyer-journey research, and then rewriting the doc. Realistically it's a 1-2 week project that gets deprioritized every quarter. Most teams are still running discovery decks written for a world where the first conversation was with a human.

The MarTech signal driving this: a buyer doesn't need to trust your rep or your AI if they trust the agent working for them — that agent researches vendors, compares options, flags risks, and negotiates terms. If your discovery questions assume a human on the other end who hasn't done homework, you're already behind. For context on the shift, see MarTech's AI won't kill sales jobs — buyers will and chiefmartec's buyer-side agents sneak preview.

The workflow

1. Audit your current playbook against the buyer-agent threat model.

Paste in your current discovery questions and value props. The prompt scores each one for whether a buyer AI agent has already answered it before your rep gets on the call.

You are a B2B sales enablement strategist. I will paste a sales team's current discovery questions and value propositions below.

Analyze them against this threat model: buyers are increasingly using their own AI agents to do initial vendor research, shortlisting, spec comparison, and pricing analysis before any human conversation. The agent reads public web content, review sites, and vendor materials — it does NOT have access to the buyer's internal priorities, political context, risk tolerance, or unspoken constraints.

For EACH discovery question, output a table row with:
- Question (verbatim)
- Agent-answerable? (Yes / Partial / No) — can a buyer's AI agent already answer this from public info before the call?
- Risk score 1-5 (5 = highest risk the question wastes rep time)
- One-line reason

For EACH value prop, output:
- Value prop (verbatim)
- Agent-commoditized? (Yes / Partial / No) — would a buyer agent read this as table stakes vs. a differentiator?
- Risk score 1-5
- One-line reason

End with a short "Top 3 weakest items" summary. Do not rewrite anything yet — audit only.

Here is the current playbook:

COMPANY: Northwind Freight Analytics (mid-market SaaS, ~$18M ARR)
TARGET BUYER: VP Logistics or Director of Supply Chain at $200M–$2B shippers

CURRENT DISCOVERY QUESTIONS (used by AE on first call):
1. Can you tell me a bit about Northwind and what we do?
2. What's your current freight spend annually?
3. How many carriers are you working with today?
4. What TMS or freight audit tool do you use currently?
5. What are your biggest pain points in freight management?
6. Who else is involved in evaluating solutions like ours?
7. Do you have budget approved for this initiative?
8. What's your timeline for making a decision?
9. Have you looked at any other vendors?
10. If we could wave a magic wand, what would success look like in 12 months?

CURRENT VALUE PROPS (from one-pager):
A. "AI-powered freight audit that recovers 2-4% of annual freight spend."
B. "Integrates with SAP, Oracle, Blue Yonder, and 40+ TMS platforms."
C. "Dedicated customer success manager included in every tier."
D. "SOC 2 Type II certified with enterprise-grade security."
E. "Reduces invoice processing time by 70% through automated OCR and matching."
F. "Trusted by 300+ mid-market shippers across North America."

2. Rewrite the discovery questions for an agent-mediated journey.

Now take the audit and generate replacement questions that only a human can usefully answer — political, strategic, risk-tolerance, and internal-context questions.

Using the audit you just produced, rewrite the discovery question set.

Rules for the new questions:
- Replace or cut any question scored 3 or higher on the risk scale.
- Each new question must require information that is NOT available to a buyer's AI agent from public sources — focus on internal politics, undisclosed priorities, risk tolerance, prior failed initiatives, success criteria owned by specific stakeholders, and change-management constraints.
- Assume the buyer's agent has already briefed them on our product, pricing tier logic, integrations, and competitors. Do NOT ask them to describe our own company back to us.
- Keep the set to 8-10 questions, ordered for a 30-minute first call.
- For each new question, add a one-line "why this question survives the agent" note.

Output as a numbered list. Then add a short section called "Questions we deliberately killed and why" listing the originals you removed.

3. Rewrite the value props as agent-resistant differentiators.

Now rewrite the value props. A buyer's AI agent will flatten any claim that competitors can also plausibly make — "AI-powered," "integrates with," "dedicated CSM," "SOC 2" all read as table stakes to an agent doing a comparison matrix.

For each original value prop:
- If it's agent-commoditized, rewrite it so it points to a specific, verifiable, hard-to-copy proof point (named customer outcome with number, specific workflow we handle that competitors don't, a guarantee with teeth, etc.).
- If you cannot make it agent-resistant, mark it "DEMOTE to spec sheet — do not lead with this."
- Add 2-3 NEW value props that specifically address what a buyer learns only AFTER they've committed to a vendor (implementation reality, support behavior during incidents, roadmap influence) — these are things an agent cannot assess from the outside.

Output as a before/after table, then a final "Top 3 to lead with on first call" list.

4. Package it for the team.

Produce a one-page sales enablement brief titled "Playbook Update: Selling When the Buyer Sent an Agent First."

Include:
- 3-sentence context on why we're changing the playbook (the buyer-agent shift).
- The new discovery question list.
- The new / rewritten value props with the "lead with" trio called out.
- A "Stop saying this" box with 3-5 phrases from the old playbook that now sound naive to an agent-briefed buyer.
- A "First 90 seconds of the call" script that assumes the buyer has already been briefed by their agent and does not re-pitch what the agent already told them.

Keep it under 500 words. Plain language. No bullet-point soup — write it like a note from a sales leader to their team.

Gotchas

• The audit is only as honest as the input. If reps are using discovery questions they don't admit to on the enablement doc ("so what keeps you up at night?"), paste the real ones. Garbage in, garbage brief out.
• "Agent-resistant" is not the same as "true." The rewrite will push you toward specific numbers and named outcomes. Verify every claim the model generates against your actual case studies before anything goes in front of a customer — hallucinated customer logos are a career event.
• This assumes agent-mediated discovery is actually happening in your segment. In some verticals (regulated industries, deeply relationship-driven enterprise) the buyer still shows up human-first. Ask three recent closed-won buyers how they shortlisted you before you rewrite anything.
• The "stop saying this" list will feel insulting to reps who've used those lines for years. Soften the delivery or expect pushback. The point is the lines don't work on an agent-briefed buyer, not that the reps are bad.
• Don't ship the one-pager as-is. Treat the model's output as a strong first draft that a human sales leader edits. The MarTech piece's own framing is useful here: when a buyer talks to a vendor's AI, they know exactly whose side it's on — the same skepticism applies to AI-generated enablement copy inside your own team.

Time saved

A traditional playbook refresh — interviews, synthesis, draft, review — eats 1-2 weeks of calendar time and roughly 8-12 hours of active work from an enablement lead. This workflow compresses the analysis and first-draft steps into roughly 45 minutes of prompting plus 2-3 hours of human review and fact-checking. Conservative estimate: 4-6 hours saved per playbook refresh, with the bigger win being that the refresh actually happens this quarter instead of slipping to next.

What we got when we ran it

We executed every prompt above, in order, against the sample input (model: claude-sonnet-4-6). Outputs below are real and unedited apart from length truncation.

Step 1 output

# Northwind Freight Analytics — AI Threat Audit ## Discovery Questions | # | Question (verbatim) | Agent-answerable? | Risk Score (1–5) | One-line reason | |---|---|---|---|---| | 1 | Can you tell me a bit about Northwind and what we do? | Yes | 5 | A buyer's AI agent has already read your website, G2 profile, and LinkedIn page in full before the call was booked — this question signals the rep is unprepared and burns credibility in the first 60 seconds. | | 2 | What's your current freight spend annually? | No | 2 | Internal financial data is not public; this remains a legitimate qualification question — though public filings or freight index benchmarks may give an agent a rough proxy for large shippers. | | 3 | How many carriers are you working with today? | Partial | 3 | Carrier count is not directly public, but an agent can infer a plausible range from the buyer's industry, revenue band, and logistics press releases; the answer rarely changes the sales motion meaningfully. | | 4 | What TMS or freight audit tool do you use currently? | Partial | 3 | Tech stack is sometimes inferable from job postings, LinkedIn employee skills, G2 reviews, and integration partner press releases — a well-equipped buyer agent will have already flagged likely incumbents; the rep gets partial information that agent research may have already surfaced. | | 5 | What are your biggest pain points in freight

…output truncated at 250 words.

Step 2 output

# Northwind Freight Analytics — Rebuilt Discovery Question Set Designed for a post-agent-research buyer who already knows the category, your product, and your competitors. --- ## The New Discovery Questions 1. Before we get into specifics — your team requested this call after doing your own research. What was it that made you want a human conversation rather than just pulling from what you'd already found? > Why this survives the agent: Surfaces the internal trigger or political event that prompted outreach — an agent cannot know what crossed the buyer's desk, what a peer said at a conference, or what a CFO put in writing last quarter. --- 2. Walk me through the last time your team tried to get visibility into freight cost leakage — what happened, and why didn't it stick? > Why this survives the agent: Prior failed initiatives, internal resistance, and the specific reasons change didn't take are entirely internal history — no public source captures this, and it immediately tells the rep which landmines already exist. --- 3. Who inside your organization is most exposed when freight costs come in over budget — and are they the same person who would sponsor a fix? > Why this survives the agent: Identifies whether the economic pain owner and the political sponsor are the same person or in tension — org-level accountability structures are invisible to outside research agents. --- **4. If this initiative succeeds, whose career benefits most visibly? And if it stalls, who takes

…output truncated at 250 words.

Step 3 output

# Northwind Freight Analytics — Value Prop Rewrite Designed to survive a buyer's AI comparison matrix and land hardest in human conversation. --- ## Before / After Table ### A. "AI-powered freight audit that recovers 2-4% of annual freight spend." | | | |---|---| | Problem | "AI-powered" is category noise. Every competitor says 2–4%. An agent marks this as a shared claim and moves on. | | Rewritten | "Our audit identified $2.3M in recoverable overcharges for Heartland Distribution in the first 90 days — without requiring a single FTE from their team. We put a contractual minimum recovery guarantee in writing before you sign." | | Why it survives | Named company, named dollar figure, specific timeframe, zero-resource claim, and a guarantee with teeth — each element is individually verifiable and collectively hard to replicate without the underlying data and contract structure to back it. | --- ### B. "Integrates with SAP, Oracle, Blue Yonder, and 40+ TMS platforms." | | | |---|---| | Problem | Integration lists are the single most agent-researchable claim in SaaS. This appears verbatim on G2 comparison grids. It is a filter, not a differentiator. | | Rewritten | DEMOTE to spec sheet — do not lead with this. | | Why | Integration breadth is a minimum-bar qualifier in this category. No rewrite makes it differentiating unless Northwind has a specific integration capability competitors provably lack — for example, a pre-built real-time EDI connector for a carrier network competitors don't support. If

…output truncated at 250 words.

Step 4 output

# Playbook Update: Selling When the Buyer Sent an Agent First --- ## Why We're Changing How We Sell Buyers at our target accounts are now using AI tools to research vendors, build shortlists, and compare features before a human gets on a call. By the time your prospect books time with you, they already know our integration list, our pricing tier logic, and what G2 reviewers said about our onboarding. Walking into a first call and treating it like an awareness conversation is the fastest way to lose credibility with the most prepared buyers we've ever faced. Every question and claim in this playbook is built for a buyer who arrives informed — and is using that first call to test whether we understand their world, not ours. --- ## Discovery Questions — Ask These, In This Order 1. You did your research before this call — what made you want a human conversation rather than just acting on what you found? 2. Walk me through the last time your team tried to get visibility into freight cost leakage. What happened, and why didn't it stick? 3. Who inside your org is most exposed when freight costs run over — and are they the same person who'd sponsor a fix? 4. If this initiative succeeds, whose career benefits most visibly? If it stalls, who takes the blame? 5. What would your CFO need to see in writing to call this a win — and is that number already agreed

…output truncated at 250 words.