Let It Toast

Overview

Let It Toast is a voice-enabled recipe capture and planning app (local Next.js + FastAPI today; no cloud deploy yet). The Assistant owns the plan; the Researcher scrapes only open slots; the Scanner saves a private recipe from a photo. Neo4j holds recorded recipes (catalog and public), plans, and friends.

System design

• Confirm, then act. Speak or type a dish; confirm before Neo4j plan or research.
• Saved recipes first, then the web. The Assistant reads catalog and public recipes from Neo4j; the Researcher finds external pages only for slots still open.
• Three agentic surfaces. Assistant and Researcher run in-process on FastAPI; Scanner is a separate /scanner path.

Product path

What a person can do. Diagram key: ✦ AI agent · purple = actor · white = service.

---
config:
  theme: default
  themeVariables:
    background: "#ffffff"
    fontSize: "14px"
---
flowchart TB
    person[Person]
    app[Let It Toast]
    assistant["✦ Assistant"]
    researcher["✦ Researcher"]
    scanner["✦ Scanner"]
    neo[Neo4j
users, friends, recipes, ingredients *relationships]

    person -->|find and catalog 
public recipes| app
    person -->|invite friends, 
share recipes| app
    person -->|scan and save 
private recipes| app
    app --> assistant
    app --> scanner
    assistant -->|get public or 
private recipes| neo
    assistant -->|new recipes needed| researcher
    researcher -->|save public recipes| neo
    scanner -->|save private recipe| neo
    neo --> person

    classDef agent fill:#e8f0ee,stroke:#0f3d36,stroke-width:2px,color:#0d0d0d
    classDef service fill:#fafaf8,stroke:#c8c8c2,stroke-width:1px,color:#0d0d0d
    classDef actor fill:#f3eef8,stroke:#6b4f9a,stroke-width:2px,color:#0d0d0d
    class assistant,researcher,scanner agent
    class app,neo service
    class person actor

Server queue, slot fill, and tool detail live on Mental Model and the agent tabs.