# Team Haske

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Andrew David](https://cerebralvalley.ai/u/emperorsixpacks)
- **GitHub:** http://github.com/emperorsixpacks/haske
- **Demo video:** https://youtube.com/shorts/ubgSRy9Op8E
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4385

Haske is a mobile-first web app built to help Nigerians identify potential counterfeit, unregistered, and falsified products before they buy them.

Counterfeit products are not always easy to spot. They can carry convincing packaging, imitate trusted brands, and even display fake or stolen NAFDAC registration numbers. For consumers, checking whether a product is properly registered or has been flagged by regulators often means navigating multiple platforms, manually searching records, and knowing what to look for. Haske makes that process simpler.

With Haske, users can point their phone camera at a product label and get its NAFDAC registration number (NRN), batch number, and expiry date read automatically. The app then checks the extracted details against NAFDAC's Greenbook, NAPAMS, and a curated database of public NAFDAC safety alerts.

The important part is how Haske handles verification. AI helps read product labels, but it doesn't decide whether a product is genuine. A deterministic verification engine evaluates the available records and determines what can actually be established.

When a product's details match an official registry, Haske reports that they are consistent with the available records, without claiming this proves authenticity. When a product matches a public NAFDAC alert, users get the source and date. If information is missing, inconsistent or unavailable, Haske returns a clear "cannot confirm" rather than guessing.

For suspicious cases, an optional investigator agent can perform additional checks through controlled tool calls. Its findings are strictly validated against actual results to prevent fabricated claims or misleading conclusions.

Haske also supports offline access to previously saved checks. Product images are processed without being stored, and no personal data is collected or retained.

The goal is simple: make product verification accessible to everyone, help consumers recognise potential warning signs and make better-informed decisions about what they buy and use.

## More from Meta Global AI Developer Hackathon

- [ShopFloor AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4382)
- [ARK 305 APP](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4383)
- [Tabrez Sait](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4384)
- [Hasami](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4386)
- [Hazmat](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4387)
- [Maple Syrup](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4388)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4385. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt