Where demand lives, where supply can meet it

Internal brief: where the demand actually concentrates

The lead. The corpus collapses into two tasks that recur far more than the company count suggests. One is *"place deformable/fragile food into a compartmented tray to a presentation spec"* — most of our 110 soft-body cases, and a task we can build now (works_in_demos, ~3mo). The other is *"de-nest fragile filled glass and assemble it into a device"* — the pharma fill-finish/autoinjector cluster, which carries the strong demand and the recall-grade failure cost, but needs primitives we can't yet build. The decision that matters: whether to pull rigid_insertion and press_fitting forward off their 6–12-month horizon to reach that second cluster, or ride the buildable food wedge for breadth.

What's jumping out:

• The most-reused task in the whole dossier is meal-tray plating. The *same* deformable-food-into-tray task appears at all 10 airline caterers (dnata, DO&CO, EKFC, Flying Food, Gate Gourmet, LSG, Newrest, SATS, Servair, TajSATS), across fresh-prepared (Bakkavor, Greencore, Hearthside, Premium Brands, Fresh & Ready) and contract dining (Aramark, Sodexo, Compass). One capability, 30+ buyers, already in demos. This is the breadth play.
• The build-next gap is the autoinjector. *"Insert prefilled glass syringe into autoinjector, snap/press housing closed"* is described in near-identical terms by Bayer, GSK, Merck, Lilly, Pfizer, Takeda, AbbVie, Roche, Sanofi, Novartis, AstraZeneca and J&J — with its twin, *fragile-glass-vial de-nesting*, in literally every pharma entry. This is the densest high-tactile + strong-demand pocket, and it sits on rigid_insertion + press_fitting (early_experiments). snap_fit is already in demos, so we're partway there.
• Deep vs. wide. sterile_fill_finish is deep — 4 companies, 6 strong, and the *only* two validated signals we hold. cell_gene_therapy is small but uniformly high-tactile and surfaces a third distinct recurring task: sterile tubing/connector mating (AZ, BMS, Novartis). contract_dining is wide-but-shallow — 10 companies, mostly weak and inferred, no named programs.
• Honesty, plainly: of 149 cases only 2 are customer-validated, both Lilly, and both are the *low-tactile, forgiving* tasks (cooler-box depal, tub handling) — not the high-tactile core our whole thesis rests on.

What to go confirm:

• Lilly's door is open on tub/cooler handling — will that same relationship confirm appetite for the high-tactile syringe-into-autoinjector step we scored "strong" but never validated?
• Is the autoinjector task truly one reusable part across the 12 pharma names, or does prefilled-syringe shoulder-geometry variability fragment it per customer?
• In food, which of the named programs (gategroup/ABB, SATS Jurong, Newrest Montreal, Aramark RoboEatz) are displaceable vs. already served?
• Will pharma actually buy a flexible cell where dedicated high-speed assembly already exists?