← Back to journey

Dishclosure: Founder and Product Lead

An operator-first investigation that revealed the real constraint wasn’t UX or adoption, but broken upstream data lineage.

Zero-to-OneData LineageSchema DesignEcosystem DiagnosisGo/No-Go Decision
Dishclosure allergen knowledge graph

Challenge

  • Food allergy mistakes carry medical, financial, and legal consequences
  • Restaurants face workflow overhead and liability if allergen info is wrong or incomplete
  • Allergen data is fragmented across suppliers, inventory systems, recipes, and menus
  • No reliable end-to-end source of truth → any diner-facing UX risks being untrustworthy
  • Build trust as infrastructure, not as messaging

Role

  • Founder and Product Lead (end-to-end ownership)
  • Market research and ecosystem analysis across supplier → operator → platform → diner
  • Designed the data model and operator workflow (schema enforcement)
  • Prototyped and validated product concepts on both operator and diner sides
  • Made the go/no-go call based on safety and infrastructure constraints

Approach & Decisions

systems investigation → source-of-truth analysis → schema enforcement → ecosystem constraint discovery → stop decision

Anchored on safety and trust
Rejected probabilistic inference/crowdsourcing/“AI labeling” without authoritative data — trust had to be engineered.
  • A 1% error rate is unacceptable in high-stakes contexts
Started with operators (not diners)
Focused on where allergen information is created and maintained, designing for existing restaurant workflows and constraints.
  • Fit into POS/inventory/recipe/menu stacks
  • Segmented operators by adoption and motivation
Designed the data model to surface reality
Built a structured schema connecting ingredients → packaged goods → recipes → menu items to make ambiguity visible.
  • Normalized inputs via an operator portal
  • Discovered supplier-level data inconsistency as a blocker
Traced the ecosystem end-to-end
Mapped supplier → operator → platform → diner flows and prototyped integrations to find where contracts broke.
  • Constraint was upstream infrastructure outside a startup’s control
Made the call to stop
Chose to pause rather than ship an unsafe or misleading product given missing supplier-level standardization.
  • Built MVPs and secured pilot interest, but reliability required ecosystem change

Outcomes

  • Validated the true constraint: broken upstream data lineage, not diner UX
  • Produced a structured allergen schema + operator workflow prototype to test feasibility
  • Mapped ecosystem contracts and identified where data reliability collapses
  • Made a principled stop decision to avoid scaling on unsafe inputs

Learnings

  • High-stakes consumer problems demand infrastructure-level solutions
  • Trust is an ecosystem property, not a UI feature
  • Operators don’t resist solutions; they resist risk and overhead
  • Knowing when to stop is part of the job

Artifacts

View all ↗
  • Product Strategy
    Market research, competitive analysis, and ICP definition.
    Open ↗
  • Data Entity-Relationship Diagram(ERD)
    Entity relationships and schema foundations for allergen lineage.
    Open ↗
  • UX Prototype
    Operator-first workflow prototype for input and validation.
    Open ↗
  • Pilot Program Agreement
    Drafting constraints, liability considerations, and policy framing.
    Open ↗
  • Technical Design Considerations
    System constraints, integration assumptions, and reliability tradeoffs.
    Open ↗