burnt: food-distribution agents for sales, procurement, and credit control

burnt is built around food distribution’s exceptions, not a clean spreadsheet version of them.

What it does

burnt describes itself as an AI platform for food distributors (YC profile). Its public product pages break the work into sales and operations, procurement, and credit control (company product page). The company says its agents validate orders, monitor inventory, communicate with suppliers, and make collection calls.

That is a coherent buyer. Food distribution is full of relationships, shortcuts, inventory exceptions, supplier communication, and old systems. A product that understands those workflows has a different starting point from a generic ERP copilot.

Fact	What the public source says
Buyer	Food distributors and teams operating the food-supply workflow
Workflows	Sales and operations, procurement, credit control
Product shape	Start with one agent or use a coordinated platform
Public claims	The site publishes accuracy, admin-time, and profit-uplift claims; they are company-reported

Why it is interesting

The strongest part is the vertical specificity. burnt does not say “automate back office” and stop. It names order validation, purchasing decisions, supplier communication, invoice collection, and the coordination problem between agents.

The buyer can therefore ask a better question: which workflow is already painful enough to automate, and which exception patterns would break the product? Procurement may be the first wedge for one distributor. Credit control may be the first wedge for another. The useful choice is visible in the workflow, not the category label.

The founders’ background is relevant. The About page says JJ is a fourth-generation food entrepreneur who started on shrimp-factory floors, while Rhea comes from three generations of spice traders and grew up around a restaurant group. That is a documented operating connection to the industry, not proof that the agents work.

What I would resolve

The headline numbers need discipline. The site presents 99.99% accuracy, 80% less admin time, and 20% profit uplift. Those are not independent tests in the source packet. Ask for the unit, baseline, period, workflow, and customer denominator before using them.

The other question is coordination. If sales, procurement, and credit agents share context, what happens when their decisions disagree? Who approves a purchase, a customer credit action, or an exception? That is where the product’s “whole platform” promise either earns its place or becomes another dashboard.

Short version: burnt is a fit for a food distributor that wants to automate a known operational queue and can explain its exceptions. It is not a generic AI employee for every distributor on day one.

Sources checked — 2026-09-19

Cohort context

burnt is listed in Summer 2025. In our 2026-09-18 directory snapshot, 112 of 166 listed companies in that cohort have YC’s primary industry label B2B (67.5%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.

Public website snapshot

Observed 2026-09-19T16:18:30.348Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.

Signal	Homepage observation
Product description metadata	Observed
Canonical link	Observed
H1 or H2 heading	Observed
Typed structured data	Observed
Docs/developer link	Not observed in this response
Pricing link	Not observed in this response
llms.txt link	Not observed in this response
Markdown alternate	Not observed in this response

Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.