SubAlert

Inspiration

Every month, thousands of small business owners discover too late that their supplier raised prices not because they were notified, but because they noticed the difference on their next invoice. A café owner in Mumbai overpaid for coffee beans for 3 months before she realized the price had jumped 23%. A print shop in London ran out of paper mid-order because their supplier went out of stock with zero warning. Large enterprises have dedicated procurement teams and software costing $50,000+ per year to monitor this. Small businesses have nothing except WhatsApp chats and spreadsheets.

What it does

SubAlert is an AI-powered supplier monitoring and intelligence platform built specifically for small businesses. It watches every supplier 24/7 and takes action before problems hit.

11 features across the full supplier intelligence lifecycle:

• 🔴 Live Price Radar — crawls every supplier daily, fires instant alerts. When prices change, the exact monthly cost impact is calculated
• 🤖 AI Switch Advisor — when a spike is detected, instantly finds 3 cheaper alternatives with a pre-filled contact email ready in one tap
• 📊 Supplier Health Score — a single 0–100 badge per supplier combining price stability, stock reliability, MOQ consistency, and risk signals, updated every 24 hours
• 📅 Market Pulse (Monday Report) — every Monday at 8 AM, a proactive AI brief with early warning signals before any price change is confirmed
• 💬 WhatsApp Alert Delivery — critical alerts sent directly to the owner's WhatsApp, because that's where small business owners actually are
• 🌍 Community Savings Feed — anonymised pricing data from 800+ businesses showing if you're overpaying vs similar businesses in your city
• ⚡ Supply Chain Stress Test — simulates worst-case scenarios and gives a Resilience Score with specific fix actions
• 📋 Contract Expiry Tracker — tracks supplier deals and generates an AI. Negotiation brief 7 days before expiry
• 💰 Budget Forecast Engine — projects supply costs 12 months forward with a "switch now vs stay" comparison chart
• ⚠️ Supplier Risk Score — AI-assessed fraud and reliability risk per supplier using public signals and community reports
• 👥 Team Collaboration — invite staff with role-based access so the whole team stays informed

How we built it

SubAlert was built entirely on MeDo through 14 structured conversations, zero manual coding, and zero external developers. We structured each conversation to build one major capability at a time, with each prompt explicitly referencing the database tables and workflows created in the previous session. MeDo handled the full stack: React 18 frontend, Supabase PostgreSQL backend, authentication, Edge Functions, and one-click deployment. Every external service runs on free tiers — Twilio trial, WhatsApp Business API free tier, and ScraperAPI free tier. Total cost to build and deploy: $0.

Challenges we ran into

1. Making alerts feel intelligent, not noisy The hardest design problem was routing; sending every single change to WhatsApp would feel like spam. We had to build a severity classification system where only CRITICAL alerts (price spike >15%, stock-out, contract expiry in 7 days) trigger WhatsApp, while lower severity goes to email or in-app only.

2. The Market Pulse early warning system Detecting trends before a price change is confirmed requires careful prompt engineering with MeDo's LLM. We needed it to reason about patterns across 90 days of price history — not just report the last data point. Getting the language right so the output felt like a genuine intelligence brief took multiple iterations.

3. Community data without compromising privacy The Community Savings Feed needed to feel useful without exposing any individual business's data. We designed the aggregation layer so that all community contributions are anonymised at the point of entry, no business name, no location specifics, only category and price range.

4. Demo Mode that judges can trust We needed judges to experience the full app in 60 seconds with no setup. Building 3 realistic demo personas (café in Mumbai, print shop in London, salon in Lagos) with complete pre-seeded data, including 90 days of price history, fired alerts, stress test results, and verified WhatsApp numbers across all 12 database tables, was the most complex single prompt in the entire project.

Accomplishments that we're proud of

• Market Pulse works. A fully autonomous proactive intelligence system that tells business owners what is about to happen, not just what already did. No small business tool has ever offered this.

• 11 features, one coherent product. Each feature connects to the others, and an alert fires, the Switch Advisor activates, the budget forecast updates, and the community feed surfaces alternatives. It feels like a platform, not a collection of widgets.

• MeDo used at full depth. 7 MeDo capabilities used meaningfully across the product, not as demos, but as the actual backbone of every feature.

• Three business types, one platform. The same app works naturally for a café in Mumbai, a print shop in London, and a salon in Lagos because the supplier monitoring problem is universal.

• $0 total cost. A production-ready B2B SaaS platform with real-time monitoring, AI intelligence, WhatsApp delivery, and community network effects built for free.

What we learned

MeDo is most powerful when you treat each conversation as a building block. The biggest lesson was structuring prompts so every new conversation explicitly referenced what the previous one built: database table names, workflow names, and connector names. When we did this well, MeDo generated exactly what we needed in one pass. When we were vague, we got vague results.

The problem statement is the product. Writing "your suppliers just raised prices, did you know?" before writing a single prompt made every feature decision obvious. We never debated whether to add a feature; we just asked: "Does this help the business owner know before it hurts?" If yes, build it. If no, skip it.

Proactive beats reactive always. Every other supplier tool tells you after something has happened. Market Pulse tells you before. That one insight changed the entire positioning of the product from an alert app to an intelligence platform.

What's next for SubAlert

• Mobile app — WhatsApp is great, but a native iOS/Android app with Push notifications is the real end state for small business owners who run their business from their phone

• Supplier onboarding via invoice scan — upload any invoice PDF, and SubAlert automatically adds all suppliers and products already partially built, needs refinement

• Predictive pricing AI not just "prices changed" but "prices will change in the next 2 weeks based on seasonal patterns and commodity trends", genuine forecasting, not just monitoring

• Regional expansion — the community pricing data becomes significantly more valuable with more users. Target: India, Nigeria, and the UK small business communities first, where the demo personas already show strong product fit

• API for accounting software — connect SubAlert to QuickBooks or Tally, so supply cost changes automatically update the business's financial projections in real time

Built With

• medo