Measuring AI Fluency in Sales Engineering

Ask ten Sales Engineers how they use AI and eight will say they use it for research and prep. That’s confirmation everyone knows they’re supposed to – not a read on fluency. Where in the deal motion? Which outputs changed? Did demo conversion go up, or did the same work just happen faster? Nobody tracks it, so nobody knows.

Self-reports on AI use skew high because the social pressure runs one direction: nobody wants to be the person who isn’t keeping up. Managers hear the team is fine, nothing concrete gets examined, and the gap only becomes visible when a competitor shows up to a deal with preparation that yours didn’t match.

The rubric runs a structured interview across four dimensions – Mindset, Strategy, Building, and Accountability – and scores each one on a 1-5 scale. As a conversation, it asks one question at a time and probes for a real example before forming any judgment. Scores stay hidden until the end so answers don’t aim at a tier. If your org has MCP integrations available – Slack, Google Docs, Notion, customer email – the tool can sweep for evidence before the interview starts, surfacing the shared prompt library or the AI-drafted brief the person didn’t think to mention. Either path feeds the same rubric.

The four dimensions

Mindset

The first question is whether AI actually changed the SE job or just made the same work go faster. Time savings is easy to claim and hard to verify; changed outcomes are specific and traceable. A Capable score requires something concrete: faster POV setup, better demo-to-technical-win conversion, less time on documentation. An Adoptive score means they think in systems across the deal and push their teammates to raise their own bar. A Transformative score means they can say what the SE role looks like in 12 months – what gets automated, what new skills matter, how the team should be structured.

How has AI changed your SE work this quarter – did a number move, or is it mostly time saved?

Strategy

Not everyone who uses AI uses it on purpose. Strategy is about deliberate placement: knowing which parts of the deal motion pay off with AI and which don’t, and having a reason for the distinction. A Capable score requires more than one-off prep – it means AI is applied across the motion, from pre-call research to RFP drafting to turning discovery notes into POV scope. An Adoptive score means they’re running pattern analysis across deals: which technical objections correlate with losses, which demo sequences close fastest. A Transformative score means they’ve decided what the team should and shouldn’t spend SE time on, and routed the repeatable work accordingly.

Across your deal motion, where do you deliberately bring AI in, and where do you deliberately keep it out?

Building

Most people who use AI start from scratch every time. A prompt that worked last week lives in a chat window they’ll never find again. Building means compounding – a refined prompt library, reusable templates, a repeatable pre-call workflow, an RFP knowledge base that produces first-pass answers teammates refine rather than rebuild. An Adoptive score requires something others actually use. Transformative means the system has feedback loops: won/lost data improving demo narratives and POV frameworks so each deal makes the next one better.

Tell me about the last thing you reused instead of rebuilding – a prompt, a template, a workflow. Does anyone else on the team use it?

Accountability

The last dimension is about what happens after the AI writes something. Capable means the person reviews and edits before anything reaches a customer – output quality is measurably better because they catch what the model gets wrong. Adoptive means they share what works: enablement sessions, shared workflows the team relies on. Transformative means they’ve enabled AEs to self-serve technical content for smaller deals, freeing SE capacity for the work that actually requires an SE.

Before any AI output reaches a customer, what happens to it? And what have you handed the team?

The scale and the read

The levels are named so a 3 and a 4 mean something:

Band first, then narrow. Before arguing 3 versus 4, answer the easier question: is this a performing area or one that needs development? The line is 3. Below it, they’re not actually doing it yet. Above it, the question is how far the impact extends – their own deals, or the team’s.

The evidence test

Every dimension scores against a real artifact. A prompt library, a shared pre-call workflow, an RFP knowledge base, a demo environment someone else uses – these are what the evidence test is looking for. A claim with nothing behind it (“I use AI for prep”) caps that dimension at 2, Emerging, regardless of how confidently it’s delivered. This is what makes the score something you can coach against rather than a number you file away.

Connected MCPs make this less dependent on memory. If the tool can see Slack, Google Docs, Notion, or customer email during the session, it finds artifacts before the interview begins. The shared doc built eight months ago and forgotten still counts as evidence. It doesn’t require anyone to remember to mention it.

Two modes

Self-assessment or manager rating, same rubric. In manager mode, the questions shift to third person and there’s one reminder: rate the practice you’ve actually observed, not what you’ve heard about. A manager who mostly hears summary updates from their AE is less equipped to score an SE than one who’s been in the deals.

Run it

For the full eight-dimension picture beyond AI fluency, How to Actually Evaluate a Sales Engineer walks the framework and the SE Performance Evaluator runs it.