TONNIC AI Agency
← Web & app development

Case study · PriceHive.app

A real consumer product, in production.

PriceHive is a community-powered grocery price-comparison platform. Shoppers find the lowest price by searching, browsing, scanning a barcode, or scanning a whole receipt. Every price is submitted by users; trust, freshness, and accuracy are scored algorithmically.

Designed, engineered, and operated end-to-end by TONNIC AI Agency. Live, with real users, continuously released.

Visit PriceHive.app →
PriceHive — Community-powered grocery prices. Compare. Save. Share.

Live scale

In production today.

0
Prices submitted
0
Distinct products tracked
0
Stores in the directory
0
Cities covered

Discovery & search

How shoppers find prices.

  • Product search with autocomplete (name, brand, synonym).
  • Category browsing — 10+ top-level categories with deep sub-trees (Fresh Vegetables alone covers 272 product types).
  • City-based browsing across featured North American cities.
  • Store locator with geolocation — nearby stores ranked by distance.
  • Per-product price history and per-store price listings.
  • Public leaderboard recognising top community contributors.

Price submission

Three ways to add a price.

  • Manual entry

    Name, brand, store, price, quantity, unit. Optional photo evidence as proof.

  • Barcode scan

    Phone camera scans the product; auto-fills details from Open Food Facts when a match exists.

  • AI receipt OCR

    User scans a receipt; the system extracts every line item, the user confirms, all prices submit in bulk.

Automatic validation on every submission

  • Sanity caps on price and quantity.
  • Outlier detection — flags prices that deviate more than 50% from the median.
  • Per-tier rate limiting (submissions per hour scale with user tier).
  • Self-confirmation prevented at the system level.

Trust & confidence scoring

Every price scored, every day.

  • Every price carries a 0–5 star confidence rating, recomputed daily.
  • Freshness state on every price: Fresh (≤2 days) · Recent (≤7 days) · Stale (>7 days).
  • Tier-weighted community confirmations (New 0.5× → Regular 1.0× → Trusted 1.5× → PowerUser 2.0×).
  • Time decay — newer confirmations count more than older ones.
  • Community flagging — validated flags reduce confidence and penalise the submitter.

Reputation & gamification

Trust is earned, not declared.

Four user tiers

  1. New
  2. Regular (≥50 rep)
  3. Trusted (≥500 rep)
  4. PowerUser (≥2,000 rep)

Five collectible badges

  • Scout— first 10 prices submitted
  • Hunter— 50 verified prices
  • TrendSetter— 10 first-to-report sales
  • Investigator— 100 confirmations given
  • Hero— 95%+ accuracy across 50+ submissions

Smart shopping lists

Four ways to optimise a route.

  • Best Overall

    Cheapest store for each item independently.

  • Single Store

    One store where the full list costs least.

  • Multi-Store

    Smart split across N stores with configurable max stores and minimum items per store.

  • Convenience

    Nearest store, ignoring price. Distance constraints honoured.

Behind the scenes

Two engines users feel, operators see.

Daily algorithmic recalculation

Every confidence score and freshness state across the full price database is refreshed automatically each day, with no manual intervention. Trust never goes stale.

AI-driven receipt OCR pipeline

Multi-provider evaluation across leading vision models, tuned for cost-per-receipt against extraction accuracy. The pipeline picks the right model per receipt.

Editorial layer

More than a price database.

PriceHive isn’t only a tool. It’s a content destination, with sales surfaced where shoppers find them and editorial published into the same product surface.

  • Sales feed

    Time-bound store sales surfaced on product and category pages. Shoppers find what is on sale where they are already looking, not in a separate destination.

  • Blog

    Articles, how-to guides, and community spotlights, all published into the same product surface and indexed alongside prices and stores.

Operator tooling

Admin tools, fully built.

The case study isn’t only the consumer app. PriceHive ships with the full operator side. TONNIC built every queue, dashboard, and curation control alongside the user-facing product, not as an afterthought.

  • Flagged-price queue

    Community-flagged prices land in a moderator queue with the surrounding context needed to act, deduplicated against the live database.

  • User management

    Tier promotions, contribution audits, and moderation actions across the user base. Reputation maintenance, not just sign-up gating.

  • Store directory curation

    Add, edit, and deduplicate stores across cities. The directory stays clean as new locations come in from the field.

  • System stats dashboard

    Usage, contributions, growth, and health metrics at a glance, so operators can see at any moment whether the product is breathing.

Built by TONNIC

The engineering work.

  • Multi-platform UI: search, browse, scan, list management, profile pages.
  • Camera-based barcode scanning with Open Food Facts integration.
  • Multi-provider AI receipt OCR pipeline with cost-vs-accuracy tuning across leading vision models.
  • Daily algorithmic confidence + freshness recalculation across the entire price database.
  • Community trust scoring with tier-weighting and time decay.
  • Reputation system, badge collection, public leaderboards.
  • Multi-mode shopping list optimisation with distance constraints.
  • Geolocation-aware store locator.
  • Full admin and moderation tooling: flagged-price queue, user management, store directory curation, system stats dashboard.
  • Editorial layer: sales feed, blog content.
  • Continuous releases — the app ships updates regularly.

How this maps to TONNIC services

What we built for ourselves, we build for clients.

Every technical pattern in PriceHive is something TONNIC ships for client projects, too.

Want one of these for your business?

We build software end-to-end with the same engineers who ship and operate PriceHive. Tell us what you have in mind.

1-888-442-BUZZ