💡 Inspiration

Cooking should be joyful but for millions of people with executive function challenges, chronic pain, and physical disabilities, most recipe apps make it harder, not easier. We all have struggled with overwhelming recipe layouts, complex multi-step instructions, and filters that didn't account for our real needs: "I'm exhausted today, what can I actually cook?" or "I can't use a knife right now."

We realized that accessibility can't be an afterthought. So we built a recipe app that centers neurodivergent and disabled cooks from day one not as an edge case, but as the design foundation.

🧠 What It Does

Simmer adapts recipes to your brain and body in real-time:

  • Natural Language Search: Tell it "I'm tired and want something warm" instead of clicking through endless dropdown menus
  • Hard Safety Constraints: Recipes that violate your allergens, excluded ingredients, or physical limitations are always blocked, no exceptions
  • "Cooking Mode" View: See one cooking step at a time with Back/Next navigation, reducing cognitive overload and making recipes actually followable
  • Accessible Recipe Step Timer: Stay focused during timed recipe steps by using the on-screen preloaded timers. Set your preferred alert method(s):
    • preset sound/uploaded audio file + volume control
    • visual alarm that flashes your screen until it's dismissed
  • Adaptive Matching: Your profile tracks persistent needs (dexterity limits, pre-cut ingredient preferences, appliance access), and optional "state chips" let you filter by today's energy level

🛠 How We Built It

Tech Stack

  • Frontend: React + TailwindCSS with responsive design and PWA capability (install Simmer to behave like a standalone app). Shadcn/ui and Lucide React components for an ARIA-friendly, screen reader optimized UI.
  • Storage: IndexedDB (via idb) for offline-first recipe persistence
  • AI/Validation: OpenAI gpt-5-nano model with Zod schema validation for type-safe recipe parsing
  • Recipe Processing: LLM-powered extraction agent (using gpt-5-nano) that mines recipe preambles for tips, substitutions, and technique notes
  • Recommendation Engine: Rule-based transparent ranking with hard filter phase (safety constraints) + soft scoring phase (energy/mood/time fit)
  • Data: Curated recipe dataset with explicit disability-aware metadata (knife-required, cleanup load, cognitive complexity, appliance compatibility)

Key Technical Decisions

  1. Profile-first architecture: Persistent disability support settings (temporary/chronic/permanent) saved to user profiles, not session-based preferences
  2. Transparent AI: We use LLM for natural language input parsing, but recommendations are rule-based and explainable (not black-box)
  3. Sequential "Cooking Mode" transformation: Recipe steps restructured for one-step-at-a-time display with contextual tips mapped from preamble content
  4. Progressive Web App (PWA): Feel trapped by the browser? Install directly to phone or computer for a distraction-free native app experience. Works offline, opens from your home screen, and updates automatically, no app store needed!

🎯 Why This Matters

15% of the global population lives with disabilities. Millions more have ADHD, chronic pain, or temporary limitations (injuries, pregnancy, aging) yet most recipe apps are designed for able-bodied, neurotypical users who don't struggle with executive function.

We built this app to prove that inclusive design isn't just ethical, it's better design. Through venting the most marginalized users you create tools that work better for everyone.

Special thanks to Taylor Arndt and Jeffrey Bishop for their accessibility agents, which we used to perform audits of our app. Check out accessibility-agents

Built With

  • openai
  • pwa
  • react
  • tailwindcss
  • vercel
  • vite
  • zod
Share this project:

Updates