Why You Need a Former Sales Engineer on Your Vendor's "New AI Features" Call

I sat in on two vendor demos this week. Different platforms, same client, same problem she didn't know she had.

Both demos were polished. The features looked impressive. The sales reps were smooth. Both platforms made a big deal about their new "AI capabilities" — and my client loved them. She was ready to pick one and move forward.

Then I started asking questions.

Not about features. Not about AI. About what happens after the demo ends.

I knew exactly what to listen for — because I used to be the guy on the other side of the screen.

I Know This Playbook. I Wrote It.

I spent years as a sales engineer. I did these demos. I knew exactly what to say, what to skip, and how to read the room to see where to steer next and what to avoid. Everything in a sales demo is designed to close the deal.

Did I answer questions truthfully? Yes. Always. But that doesn't mean I didn't follow a truthful answer with a carefully placed statement to steer the conversation away from the uncomfortable follow-up. That's not deception — it's professional selling. And every good SE does it.

Here's how it works from the inside:

• You control the narrative. The demo follows a script designed to highlight strengths and glide past limitations. If the prospect doesn't ask, you don't volunteer.
• You read the room. When a buyer leans in, you double down. When they look confused or hesitant, you pivot to something safe — usually a flashy feature or a customer success story.
• You answer and redirect. "Yes, we support that — and let me show you something our customers really love..." That pivot is deliberate. The answer was honest. But the redirect ensures nobody digs deeper.
• You know the landmines. Every platform has them — integration gaps, pricing cliffs, features that work great in demo environments and break in production. A good SE knows every one and has a bridge statement ready.

This isn't a criticism of salespeople. It's how the game works. But it means the person buying the software is always at an information disadvantage — and most business owners don't even realize it.

My client certainly didn't. She's sharp. She runs a successful business in a competitive industry. But she's done maybe two or three software evaluations in her career. The sales reps on those calls do two or three a day.

That gap is where expensive mistakes happen.

The Questions That Change the Conversation

On both calls, I waited for the vendor to finish showing a feature, then asked the question underneath the feature. Not hostile — just specific. The kind of question a former SE knows will land, because it's the one we were trained to redirect away from.

Here's what I mean:

On Integration

"You showed a beautiful dashboard. Where does this data come from? Does this pull from our existing CRM in real-time, or are we re-entering data? What's the API — REST? GraphQL? Is there rate limiting? What authentication model?"

Why this matters: Every vendor says "yes, we integrate." The word "integrate" does a lot of heavy lifting. It can mean native two-way sync, a one-way import that requires a manual trigger, or a Zapier connector someone built on a Friday afternoon. If you don't ask how it integrates, you won't find out until you're three months in and your team is re-keying data by hand.

On Data Portability

"If we decide to leave in 18 months, what does the export look like? Can we get a full data dump in a standard format? Who owns the data models we build inside the platform?"

Why this matters: This is the question vendors least want to answer, because the honest answer often exposes lock-in. You might get your raw data out in a CSV — but the workflows, automations, and custom configurations you spent months building? Those usually stay behind. You're starting from scratch with the next vendor.

On Infrastructure Compatibility

"We use SharePoint for document management and Google Workspace for collaboration. Does this platform have native connectors, or are we looking at middleware?"

Why this matters: Your business already runs on specific infrastructure. The platform you're evaluating either reaches into those systems natively, or it doesn't. The difference between a native connector with two-way sync and a one-way import with manual triggers is the difference between a connected system and an expensive clipboard.

On Scaling Costs

"Your pricing page shows per-user tiers. What happens when we go from 15 to 40 users? Do the automation limits scale, or do we hit a ceiling and have to upgrade to enterprise?"

Why this matters: I call some pricing models "lobster traps" — easy to get into, expensive to grow within, painful to leave. The per-seat rate at 10 users looks reasonable. But some platforms gate critical features behind higher tiers, add overage fees on automations, or force an enterprise upgrade the moment you need capacity. Get the real number at 3x your current scale before you sign.

On Roadmap and AI Dependency

"Where does AI fit in your product roadmap? Are the AI features built in-house or are you wrapping someone else's API? If OpenAI changes their pricing, how does that affect what we pay?"

Why this matters: If the platform's AI features depend entirely on a third-party model provider, you're exposed to the same pricing correction risk that hits developer-built agents. The dependency is just hidden under a subscription fee. This question makes vendors pause — and the length of that pause tells you something.

The "AI-Powered" Trap

This one deserves its own section, because it's the hottest button in enterprise software right now — and the one most likely to cost you money.

Both vendors on these calls led with their new AI capabilities. "AI-powered workflows." "Intelligent document processing." "Built-in AI assistant." My client's eyes lit up. She loved it.

But she didn't know what to ask. And the vendors were counting on that.

Here's the reality: every software company on the planet is racing to slap "AI" on their product right now. Some have built genuine, deeply integrated intelligence. Others have bolted on a ChatGPT wrapper and called it a feature. From the demo screen, they look identical.

The questions that separate real AI from marketing AI:

• "Is the AI trained on our data or generic data?" A generic model gives generic answers. If it hasn't learned your industry, your documents, your processes — it's a parlor trick, not a business tool.
• "What happens to our data when it goes through the AI?" Is it stored? Used for training? Sent to a third-party model provider? This matters for compliance, IP protection, and competitive sensitivity.
• "Can we see the AI work on a real scenario — not a canned demo?" Prepared demos use pre-loaded data designed to produce perfect outputs. Ask them to run it against something messy and real. Watch what happens.
• "What's the AI's error rate, and how do you handle it?" Every AI system hallucinates or makes mistakes. The question isn't whether it's perfect — it's whether the platform has guardrails, human review steps, and correction mechanisms.
• "If the underlying AI model changes or the provider raises prices, how does that affect us?" If their AI depends on OpenAI or Anthropic, you're one pricing change away from a surprise on your invoice — or a feature that quietly degrades.

My client wouldn't have known to ask any of these. That's not a gap in her knowledge — it's a gap in the sales process. The vendor's job is to sell. Your job is to decide. Those are fundamentally different objectives, and the AI hype cycle is making the gap wider, not smaller.

The 10 Questions Every Business Owner Should Ask

You don't need a technical background to ask these. You just need to know they exist. Print this list. Bring it to your next vendor demo.

1. What does your API actually support?
Read-only or read/write? Rate limits? Authentication? If the vendor can't answer clearly, that's your answer.

2. How does this connect to our existing systems?
Not "can it" — how does it, specifically? Native connectors, middleware, or manual data entry?

3. What happens to our data if we leave?
Full export in standard formats? Or are we leaving behind configurations, workflows, and relationships that took months to build?

4. What's the total cost at 3x our current user count?
Not the per-seat rate — the actual total. Include overage fees, premium feature gates, and support tiers.

5. Where do the AI features come from?
Built in-house? Wrapping OpenAI? Using open-source models? This tells you about cost stability and control.

6. What's your uptime SLA and who's responsible when it breaks?
Especially important if this platform becomes central to daily operations.

7. How does this handle our document management?
SharePoint, Google Drive, AWS S3 — wherever your files live. Does the platform reach in natively, or does someone have to move files manually?

8. What security certifications do you hold?
SOC 2, ISO 27001, HIPAA if relevant. Ask to see the report, not just the badge on the website.

9. What does implementation actually look like?
Timeline, resource requirements from your team, data migration scope, training. "Easy setup" is marketing — get the specifics.

10. Can we talk to a customer in our industry who's been live for more than 6 months?
New customers love everything. Six-month customers know the truth. If the vendor hesitates, that tells you something.

Why Features Don't Matter If the Foundation Is Wrong

My client's reaction after both calls was telling. She said: "I didn't even know to ask those questions."

That's not a knock on her intelligence. She's sharp — she runs a successful business in a competitive industry. But evaluating enterprise software platforms is not her job. She's done it maybe twice in her career. The vendor's sales rep does it twice a day.

This is why feature comparisons are dangerous. Two platforms can look identical in a side-by-side feature checklist and be fundamentally different in how they integrate, scale, lock you in, and handle your data. Features are the surface. The foundation is what determines whether you're still happy in 18 months.

The Case for Having a Former Sales Engineer in the Room

I didn't join those calls to sell anything. I was there to ask the questions my client didn't know to ask — about APIs, data portability, integration architecture, AI dependencies, scaling costs, and vendor lock-in.

I knew what to listen for because I spent years on the other side of those calls. I know the playbook. I know the redirects. I know when a truthful answer is being followed by a carefully placed pivot to keep you from asking the dangerous follow-up. And I know the technical questions that make sales engineers pause — because those are the ones that expose real limitations.

And when every vendor is leading with their AI feature set — which they all are now — it helps to have spent time teaching practitioners how to actually design and build AI products. As a facilitator for MIT xPRO's Designing and Building AI Products and Solutions program, I know what should be under the hood. That means I can tell when a vendor's "AI-powered" capability is real architecture and when it's a marketing layer on top of basic automation.

Afterwards, I wrote up two assessments. Not feature comparisons — operations comparisons. What integrates, what doesn't, where the lock-in risks are, which AI claims hold up under scrutiny, and what her team should be watching for after implementation.

That took me a few hours. It could save her organization six figures in switching costs, re-implementation, and lost productivity if the wrong platform gets selected.

If you're evaluating AI or operations platforms, you don't need another demo. You need someone in the room who knows the sales playbook from the inside and spent decades in the operator's chair. Someone who can separate the real capabilities from the pitch.

That's what Strategic Advisory looks like in practice. Not theory. Not a framework. A former sales engineer, operations veteran, and MIT AI facilitator who sits in on the calls and asks the questions that protect your investment before you sign anything.

Let's talk before your next vendor call.