Carryout Deal Radar
A local cheap-eats intelligence service — a free daily email and web page of the freshest carryout deals near you, ranked honestly by what actually feeds a family, built end-to-end by Mad Monkey Creative.
- Product
- Carryout Deal Radar
- Industry
- Local consumer food-tech
- Type
- Owned MMC product
About the project
Every neighborhood has great carryout deals — Tuesday soup specials, family packs, pizza-night steals. The problem is that they're buried in apps you forget, emails you skip, and signs you drive past. Nobody has time to hunt them down every night, so most families quietly overpay for dinner.
Carryout Deal Radar fixes that with a simple promise: tonight's cheapest dinner, already found. Every morning an AI pipeline reads local restaurant sources, turns messy menus into structured deals, verifies every price against the restaurant's own published text, and ranks the results by practical family value — what it costs per person, how many it actually feeds, and whether it fits a real budget. The winners go out in a free daily email and on a public deal page for each town.
What makes it different is the trust rule at its core: the ranked list is algorithmic and un-buyable. Restaurants will be able to buy clearly-labeled placement around the list, but never a position inside it. The ranking is the product, and it isn't for sale.
Project goals
- Make the nightly "what can we pick up under $40" decision a solved problem for local families.
- Build the audience asset first: a daily email list with double opt-in, one-click unsubscribe, and honest social-proof counters.
- Automate deal supply with an AI extraction pipeline that never invents a price — every published deal is verified against the restaurant's own source text.
- Keep the ranking honest by design: paid placement is labeled and physically separate, never mixed into the organic list.
- Architect for expansion: every restaurant, deal, and subscriber is scoped to a market so the next town is a configuration exercise, not a rewrite.
Services used
Application Design (SaaS). A full product: scoring engine that converts prices and portion estimates into cost-per-person math, a review workflow for every AI-extracted deal, subscriber management, and a self-serve roadmap for restaurants.
Website Design. A warm, accessible design system — WCAG 2.1 AA in light and dark — spanning the marketing page, the daily deal pages, and the transactional email templates.
Hosting. Runs on MMC-managed infrastructure with containerized services, automated database migrations, daily backups, and uptime monitoring.
Marketing. The daily email is the flagship surface: deliverability engineering (DKIM, DMARC, one-click unsubscribe), referral hooks, and cross-promotion with sister product Time To Plate.
Technologies & platforms
- Python / FastAPI with server-rendered Jinja2 pages (SEO-first)
- PostgreSQL with market-scoped multi-tenancy
- Playwright headless-browser scraping with snapshot storage
- Anthropic Claude for structured deal extraction with anti-fabrication guardrails
- ForwardEmail for authenticated daily sending (DKIM / DMARC / RFC 8058)
- Docker Compose deployment with zero-downtime migrations