One Door
One Door is an AI benefits advocate for Californians who qualify for help but get stopped by confusing rules, language barriers, documents, and phone calls.
Core trick
Computes CalFresh eligibility and benefit amounts in a deterministic rules engine, with the LLM only translating explanations and never altering digits or citations.
Description
One Door is an AI benefits advocate for Californians who qualify for help but get stopped by confusing rules, language barriers, documents, and phone calls.
Every year about 2.7 million eligible Californians receive no CalFresh (SNAP) benefits, leaving roughly $3.5 billion in already-appropriated federal food dollars unclaimed. A common cause: screeners run the federal default rule (130% of the poverty line) and never apply California's higher 200% limit, so a working family that actually qualifies is told "no."
We built a working CalFresh eligibility advocate that pairs a deterministic, cited benefit engine with Opus 4.8 for the human layer. It takes a household's situation in any language, computes a cited CalFresh verdict and dollar amount, explains it in the person's own language (every digit and citation stays byte-identical), reads pay stubs, surfaces adjacent programs, and refuses to invent a dollar figure for any program it did not fully compute.
Our demo runs the highest-stakes path: a California household that looks ineligible under federal rules but is eligible under California's. One Door flips the verdict, shows the estimated benefit ($231/month), cites every rule that got there, translates the guidance, and preps the person for their county interview, all without inventing amounts for programs it did not verify.
Core principle: use AI for the messy human parts (language, documents, fact triage, advocacy, navigation) and keep benefit math deterministic, testable, and auditable. Final verification gate: 30 passed / 0 failed, with independent-oracle accuracy of 100% on 14 hand-worked cases.
README excerpt
# One Door Advocate A conversational benefits advocate for California. You talk to it in **any language**, and it helps you understand what you likely qualify for, what to bring to your interview, and how to apply, with every dollar figure and eligibility verdict computed by a deterministic, cited engine rather than guessed by a language model. > Education and navigation, never an official determination, legal, or medical advice. Built for Claude Build Day (Cerebral Valley + Anthropic). California-first; the seed persona "Maria" is a cited composite, not a real user. --- ## The grounding contract (the whole point) The language model **orchestrates and converses; it never computes eligibility or invents a rule or a dollar amount.** Opus runs the conversation and decides what to ask, then calls tools that delegate to the deterministic engine: - `screen_eligibility` — every eligibility verdict and every monthly-benefit dollar comes from here. - `recommend_stack` — other California programs the household likely qualifies for (cited, no dollar math). - `get_rule_citation` — the exact policy paragraph and source URL behind a named rule. - `set_language` — switches the whole interface into the person's language (see below). **Every served fact is engine-computed and cited to a policy paragraph + URL, or it is flagged as general guidance.** A dollar figure or verdict that did not come from the engine is never stated. The engine itself is a policy-to-code compiler: it reads th