EXPOSED
A large slice of the sales engineer's week — RFP and RFI responses, spec sheets, configuration quotes, technical FAQ answers, competitive comparison decks, follow-up summaries — is exactly the text-and-spec work generative AI already does at usable quality, and interactive product tours are increasingly self-serve. What survives is the live part: reading a room of skeptical engineers, running a proof-of-concept on the customer's messy actual infrastructure, and being the person who says 'our product will not do that' and is believed. Modal workers in complex B2B capital equipment and enterprise software keep their seat; those whose job is mostly configuring quotes and answering standard technical questions will see headcount thin.
Mixed — a routine tier and a judgment tier. At 9, roughly half the week — RFP responses, bill-of-materials configuration, sizing calculators, standard technical objection handling — is text and spec work a model handles now, while the other half (scoping a POC against a customer's undocumented legacy stack, diagnosing why the integration failed on their network, whiteboarding architecture live with their engineers) still needs a person in the loop, which is why it sits mid-band rather than at 5 with inside sales or at 14 with field service techs.
Some physical or field component. An 8 reflects that most of the job runs from a laptop on Zoom, but the deals that matter still involve flying to a plant floor to measure clearances, hooking demo hardware to the customer's PLC or test bench, or sitting in their datacenter during a pilot cutover — travel and on-site equipment handling that is real but intermittent, not the daily uncontrolled-environment work of an installer.
No licence, no signature requirement. A 2 is accurate because no state licenses sales engineers; a PE stamp is rare and never required to quote or demo, and when a misconfigured spec causes a failure the exposure lands on the employer's contract terms and the customer's own engineering sign-off, not on the individual who built the proposal.
The human relationship is the product. At 14, the buying engineer's willingness to believe 'no, our product won't do that' is the asset — that credibility is built over multi-quarter evaluation cycles and follows the individual to their next employer — though it stops short of 18 because the account is co-owned with the AE and backed by the vendor's brand, reference customers, and support organization rather than resting on one person's name.
Meaningful discretion. An 11 fits calls like whether to commit to a custom integration in writing, which failure mode to disclose before the customer discovers it in production, and how to scope a POC so it proves the thing that actually blocks the deal — genuine discretion with revenue and reputational consequences, but decisions that get reviewed by sales management, product, and legal before they become binding.
Has AI actually changed your work?