MoveTech
Everyday internet and mobile users in Azerbaijan who receive suspicious SMS messages, links, or screenshots claiming to be from banks, telecom operators, or other organizations, and who do not have the technical knowledge to tell whether a message is real or a scam. This includes people who are less experienced with technology, such as older adults and first-time online banking users, as well as anyone who has received an urgent message like "your card has been frozen, click here to activate it" and is unsure what to do. Phishing and impersonation scams are hard to recognize because they look real. A fake message uses a well-known brand name (a bank or a mobile operator), creates urgency ("your card is frozen, act now"), and sends the victim to a look-alike domain (for example, azercell-az-secur.net) that steals personal and card data. Most people have no simple way to check such a message on their own. Technical tools for analyzing domains and phishing are built for security specialists, and searching for an organization's official contact details takes time while the message pressures the user to act immediately. As a result, users either ignore real messages or fall for fake ones. ScamGuard solves this: the user pastes a suspicious message or link, or uploads a screenshot, and AI instantly explains in simple language whether it is risky, shows a risk score and the warning signs, recommends what to do, and provides the organization's official website, phone number, and email so the user can verify it independently.
Preliminary score
- Value for the user17.5 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI16.0 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing3.5 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility5.5 / 15
Data requirements, running costs and a clear next step.
- Originality5.0 / 10
A meaningful difference from common solutions.
What the judges wrote
- Claude Opus 5.5Preliminary46/100
- Strengths
- Your problem framing is concrete and local: Azerbaijani users receiving urgent fake bank or Azercell SMS with look-alike domains such as azercell-az-secur.net, and older or first-time online banking users as the core audience. The described flow covers text, URL and screenshot input and returns a risk score, warning signs, recommended actions and official contact details, so AI is central to the result. Pairing the verdict with verified official contacts from your own database is a useful step beyond a bare 'scam or not' answer. Clear setup instructions with a sample message and expected output make the core scenario easy to reproduce.
- To improve
- Your quality testing currently consists of limitations and one expected example. Build a labelled set of real and fake Azerbaijani SMS, screenshots and URLs, report how often ScamGuard is right and wrong, and show specific misclassifications, including false alarms on genuine bank messages. Compare it with what users do today, such as calling the bank or checking the domain manually. Add per-check API cost, how you will keep and grow the organization contact database, and a concrete next step. A bank or telecom pilot would fit the Enterprise track, with a metric such as scam reports deflected or fraud losses avoided.
- GPT-6 AstraPreliminary49/100
- Strengths
- Your focus on Azerbaijani users receiving bank and telecom impersonation messages gives the product a clear, concrete use case. Pairing simple Azerbaijani explanations with official contact details from a curated database offers a practical verification path beyond an AI-generated risk score. Your explicit warnings about uncertainty and missing organisations appropriately acknowledge important limitations.
- To improve
- Your Azercell example specifies an expected result; strengthen the evidence with recorded end-to-end outputs for text, URL and screenshot inputs. Test labelled legitimate and phishing messages, reporting false positives and missed scams against manual verification, including unknown organisations and screenshot-reading failures. For the enterprise track, identify the bank or telecom team adopting this workflow and measure an outcome such as verification time or support escalations. Document per-check API costs, sensitive-data handling, contact-database maintenance and a concrete next implementation step.
The team
Osman Nezerli
developer
Metin QurbanovLead
developer