From blank quote to better question

The tab with the RFP is open. Sixty pages of requirements, a few drawings, some contradictions. The salesperson stares at a blank quote and asks the real question: Where do I even start?

I remember a review back at Tacton where a strong rep had three valid directions on the whiteboard and no confidence to pick one. He didn’t need help drafting an email. He needed someone to say, “Given these constraints, start here, then decide between A and B based on X.” That is the work in complex sales.

Right now, vendors are pitching AI Copilots like they’re the answer. Copilots are decent at text, summaries, and email. In complex B2B sales, the bottleneck is different. It’s the cognitive load of navigating product logic, constraints, and trade-offs to land on a valid, buildable solution. You don’t need a better typist. You need a safer brain on call.

CPQ isn’t about automation. It’s about correctness.

The useful mental model is simple: a generalist Copilot helps you write. A specialist Coach helps you decide. For multi-million dollar systems with strict configuration rules, safety and correctness beat fluency every time.

Designing the AI Coach

The ‘aha’ moment: the right AI for complex quoting is not a standalone brain. It’s a conversational layer that sits on top of your deterministic engine of truth — your product configuration logic, your pricing rules, your approved language. The AI is trustworthy precisely because it cannot guess. It reasons within guardrails you already test and govern.

Three principles make it work:

  • Correctness over fluency. If a suggestion isn’t guaranteed valid by your configuration rules, it doesn’t go in front of a customer.
  • Explainability drives adoption. Every recommendation needs a short “why this is valid” that a rep can repeat to a buyer. If the system can’t explain itself, sales won’t trust it.
  • Human in control. The AI proposes. Sales chooses. The system records reasoning and options so decisions are transparent and teachable.

Architecturally, the AI Coach is a thin, conversational layer on top of a deterministic CPQ core. The core holds product logic (compatibility, constraints, sizing), pricing and approvals, and a curated knowledge base (approved specs, performance tables, option descriptions). The Coach translates messy input — emails, notes, RFP lines — into structured questions to the CPQ engine. It then returns valid starting configurations, trade-offs, and language that draws only from approved content. No open internet. No freeform guessing. A test suite protects the rules so every “why” remains true after product updates.

AI doesn’t replace logic. It depends on it.

Call it hybrid intelligence. Humans define intent and negotiate outcomes. Rules ensure correctness. AI accelerates interaction. Remove any one layer and the system slows down or breaks trust.

Five coaching moments that matter

1) Starting the quote

Coaching moment: A rep asks, “Given cooling load, footprint, and local regulations, what’s a safe first configuration?” The Coach proposes a valid starting point because it’s grounded in your configuration rules. It might say, “Start with Frame M, Compressor Set 2x, Option B for filtration. This satisfies airflow and voltage. If weight is a concern, switch to Frame S and add Support Kit.” Sales still decides. The fear of the blank page is gone.

2) Translating requirements

Customers describe needs in their own language: “We need pharma-grade washdown,” “Class II Div 2,” or vague statements buried in page 37. The Coach maps that language to your defined attributes and options, flags contradictions, and asks crisp clarifying questions. It doesn’t hallucinate specs. It aligns words the customer uses with the rules you’ve already codified.

3) Writing the quote

The most variable part of quoting is the narrative. The Coach generates the technical description only for the correctly configured items, including approved option language and alternatives. It improves structure, keeps terminology consistent across regions, and outputs the right attachments. Fluency is finally useful because correctness is guaranteed underneath.

4) Finding product knowledge

Sales usually searches old quotes, pings an expert, or spelunks through SharePoint. The Coach answers product questions from your curated knowledge, not the open web: “What’s the maximum span for Beam Type L with Load X?” “Which skid fits a 2.2m ceiling?” Answers come with references to the rule or document section so they can be verified and shared with a buyer.

5) Responding faster without cutting corners

The cumulative effect is a short time-to-first-draft with higher confidence. Quotes decay like fruit on a shelf — every day, they’re a little harder to sell. The Coach reduces that decay by getting to a valid first path fast, then helping sales refine. Speed and safety, together.

AI should guide sales decisions. Not replace them.

Notice what’s not happening: the AI isn’t approving exceptions, inventing options, or suggesting illegal bundles. It stays inside your rules and flags where human judgment is required. That’s why legal, engineering, and product can sleep at night.

What to change this quarter

Start by making the core teachable. If your product logic is brittle or duplicated, no AI will save you. Clean logic makes AI useful. Messy logic makes AI dangerous.

  • Audit your rules for explainability. Pick 10 past deals. For each key choice, can you point to a single rule or rationale? If not, refactor until you can.
  • Map customer language to attributes. Take the top 30 phrases from recent RFQs and align them to named fields in your model. Add tests so the mapping doesn’t drift.
  • Curate your knowledge base. Move approved descriptions, specs, and option texts into a single source. Tag them to the product model so the Coach can cite them.
  • Wire a test harness. Treat your product logic like code. Version it, test it, and promote changes safely. AI on sand is still sand.
  • Pilot the Coach in a sandbox. Feed it three real opportunities. Measure two things: time to first valid configuration and the number of expert interrupts. Adoption is the only metric that matters.

If a vendor shows you a Copilot that writes beautiful emails, ask them to handle a contradictory requirement with regional compliance and legacy constraints. If they can’t show the guardrails, it’s not built for your world.

One more reality check. You may hear, “We can’t do this until pricing is perfect.” Pricing improves with usage and feedback, not with another year of spreadsheets. Progress beats perfection. Get the configuration right and let pricing learn.

The hardest part of CPQ has always been modeling the product so logic is composable, readable, and testable. Do that, and an AI Coach becomes like a GPS for complex sales. You set the destination; it keeps you on the valid roads, explains why a turn is blocked, and gets you to a correct quote without detours.

Don’t ask vendors how their AI helps sales write. Ask how it helps sales decide.