Veris

• 

  Logo

• 

  Architecture Flow

Inspiration

Leading digital literacy workshops in senior centers, I worked alongside many older adults who had firsthand experience with cybersecurity scams, sometimes losing tens of thousands of dollars after a single email or malicious website. More than money, many felt embarrassed, confused, and afraid to use the Internet.

Yet, existing anti-virus and security tools fail the most-targeted group: seniors. Interfaces use tiny fonts, technical jargon, and confusing dashboards, so seniors either ignore warnings or get frightened away from useful online services.

Veris was born to replace fear with confidence: a one-click, readable, trustworthy assistant that explains why something is risky in plain language and teaches safer habits over time.

What it does

One-click scan of selected email text, links, or attachments in Chrome.

Returns a Red / Yellow / Green verdict plus a single plain-language sentence explaining the risk (e.g., “Sender domain is paypalsecure-24.com — not paypal.com”).

Fuses signals from: Gemini Prompting (Prompt API) for contextual checks and human-friendly explanations,

Google Safe Browsing (fast, client-friendly URL reputation),

Local-first & privacy-minded: Gemini Nano runs client-side for low latency and privacy; optional hybrid fallback to Gemini Developer API via Firebase AI Logic only with user consent.

Personalized micro-lessons: local history of verdicts converts repeated patterns into tips (e.g., “Don’t trust sites that don’t start with https”).

Accessibility-first UI: large fonts, single primary action, screen-reader support, and spoken explanations.

How we built it

Chrome Extension (MV3): TypeScript + React + TailwindCSS for a high-contrast, readable UI.

Client-side AI: Gemini Prompt API for Web & Chrome Extensions — primary explainability and verdict generation.

Reputation: Google Safe Browsing for immediate URL/domain flags.

Hybrid backend: FastAPI (Python) to orchestrate heavier multimodal checks (Gemini Developer API) and optional attachment scanning.

Privacy techniques: local-first preprocessing that masks PII before remote calls; encrypted local storage for verdict history; guest mode to avoid mandatory login.

Challenges we ran into

Prompt design — tuning Gemini prompts to produce short, consistent, non-technical explanations took many iterations.

Accessible UX — designing a single-action UI that communicates risk without overwhelming or patronizing older adults required repeated user testing and copy editing.

Accomplishments that we're proud of

Working Chrome extension prototype with one-click scanning that returns clear verdicts and explanations.

Integrated Google Safe Browsing + Gemini Prompting into a single flow that prioritizes client-side checks.

Designed an accessible interface tailored for seniors (large fonts, clear colors, spoken explanation).

What we learned

Empathy drives better security design: wording, contrast, and simplicity matter more than feature density for adoption among seniors.

Explainability builds trust: users are more likely to follow a recommendation when they understand why it’s given.

Local-first models like Gemini Nano enable meaningful privacy gains and faster responses for client-side safety tools.

What's next for Veris

SMS/phone-call detection: use Twilio + Whisper to detect robocall and SMS scams in real time.

Partnerships: pre-configure Veris for public computers at senior centers, libraries, and AARP chapters.

Localization: Spanish, Mandarin, Hindi for broader reach (Translator API).

Cross-browser: Firefox & Edge versions.

Threat analytics pipeline: anonymized signals + foundation models to help link attacks and assist law enforcement (long-term).

Built With

• chrome-mv3
• fastapi
• gemini-api
• github
• google-safe-browsing
• python
• react
• tailwindcss
• typescript