EXPOSED
The core of this job — standing in a grocery aisle cooking samples, running a trade-show booth, letting shoppers touch and taste a product — is physical and social, and neither language models nor current robotics can do it. The real threat isn't a machine taking the tasks; it's marketing budgets shifting from in-person sampling to paid social, influencer content, and AI-generated product video, which shrinks the number of demo shifts booked. The screen-side portions (writing pitch scripts, logging lead cards, filing shift reports, tailoring talking points by store) are already automatable, and there is no license or liability requirement anchoring the role.
Mixed — a routine tier and a judgment tier. Handing a shopper a hot sample, reading their hesitation, and switching from the health angle to the price angle mid-sentence is not reproducible by software, but the paperwork wrapped around it — the pitch script, the per-store talking-point sheet, the end-of-shift lead count and photo of your table — is already being generated and filed automatically, which is what keeps this at 12 rather than 16.
Hands-on in uncontrolled environments. You are on your feet six to eight hours in an unstructured retail aisle: hauling your own kit, setting up a folding table, running a hot plate or toaster oven under store food-safety rules, restocking your own product, and repositioning when the store manager wants the endcap back — uncontrolled enough for 13, but it's a grocery store rather than a roof or a trench, so it stops there.
No licence, no signature requirement. Nothing licenses you to hand out cheese cubes; a food-handler card in some jurisdictions and the brand's own two-hour onboarding video is the whole gate, and if a product claim goes wrong the manufacturer's legal department answers for it, not you.
Some relationship component. Regular shoppers at your store will recognize you and some will buy because it's you asking, and brands do rebook demonstrators who move volume — but the shift is booked by an agency against a store list, the customer came for groceries, and the relationship rarely survives your reassignment to a different location, which puts it at 9 instead of 14.
Executes defined procedures on defined inputs. Your discretion is real but narrow: which shoppers to approach, when to cut a sample short, how to handle a complaint about last week's batch — all inside a brand playbook that dictates the claims you may make, the portion size, and the price point, so the consequential calls are made above you.
Has AI actually changed your work?