CraveLens

Product Introduction

1. Definition: CraveLens is a specialized Chrome browser extension that functions as an on-device visual food recognition and contextual commerce tool. It operates as a local AI agent that analyzes video frames or selected webpage regions directly on the user's machine to identify food items and subsequently interfaces with the Swiggy food delivery platform to prepare a personalized shopping cart.
2. Core Value Proposition: CraveLens exists to eliminate the friction between visual food discovery and online ordering. Its primary value is enabling impulse-driven food ordering by creating a direct, automated bridge from food content consumption (e.g., YouTube videos, Instagram Reels) to a ready-to-checkout Swiggy cart, all while maintaining user privacy through a 100% on-device AI architecture.

Main Features

1. On-Device Visual Food Detection & Identification: This is the core technical feature. CraveLens uses a multi-stage, local AI pipeline. First, a lightweight vision model acts as a real-time trigger to detect potential food items in video streams. Second, a more sophisticated local Vision-Language Model (VLM)—such as Gemini Nano, a browser-downloaded Gemma 3n, or an Ollama-hosted model—analyzes the captured frame or selected region. It identifies specific dishes, assigns a confidence score, and extracts contextual cues like cuisine and ingredients. Crucially, all image processing occurs locally; no visual data is uploaded to external servers.
2. Agentic Cart Preparation & Personalization: Upon identifying a dish, CraveLens activates a ReAct-style AI agent (a reasoning and acting framework) that functions as a "food concierge." This agent connects to Swiggy's ecosystem via the Swiggy Builders Club MCP gateway. It uses the identified dish, combined with the user's configured dietary preferences (e.g., vegetarian), allergy alerts, and order history (for address and tentative preferences) to search live menus, apply available coupons, and prepare a personalized cart with the best nearby match, including price, rating, and delivery time.
3. Multi-Source Content Integration: The extension supports automated detection on major video platforms including YouTube (videos and Shorts), Instagram Reels, and Facebook Watch. For unsupported pages or specific images, it provides a manual lasso selection tool (activated via keyboard shortcut) that allows users to capture any visible region of a webpage for local AI analysis, extending its utility beyond social media to food blogs, news articles, or recipe sites.

Problems Solved

1. Pain Point: The "Discovery-to-Order Friction" problem. When users see appealing food in digital content, the process to order it is manual and disruptive: pausing content, switching apps, searching by text (often inaccurately describing the visual item), comparing restaurants, and rebuilding the cart. This friction often kills the impulse to order.
2. Target Audience: The primary user personas are urban Indian consumers who frequently use Swiggy and consume food-centric digital content. This includes late-night scrollers, food video enthusiasts (travel vlogs, cooking shows, mukbangs), impulse buyers, and privacy-conscious tech users who are wary of cloud-based screen analysis.
3. Use Cases:
   • Watching a travel vlog on YouTube where a street food dish appears; CraveLens automatically detects it and prepares a cart for a similar dish from a local restaurant.
   • Scrolling through Instagram Reels featuring a specific type of sushi; using the lasso tool to select it and instantly finding a nearby Japanese restaurant offering it.
   • Reading a food blog about a specific pasta recipe; selecting the image to find restaurants that deliver a similar aglio e olio or carbonara.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros:
     • Uncompromised Privacy: The 100% on-device AI model is a major strength for privacy. Video frames and browsing activity never leave the user's computer.
     • Seamless User Experience: Drastically reduces steps from craving to cart, capturing impulse intent effectively.
     • Contextual Personalization: Goes beyond simple search by integrating dietary and historical preference data into the Swiggy query.
     • No Automatic Ordering: Acts as an assistant, not an autonomous agent, ensuring user control over the final purchase decision.
   • Cons / Limitations:
     • Platform Lock-in: Currently exclusive to Google Chrome with WebGPU support and tightly integrated only with Swiggy in India, limiting its market.
     • AI Model Dependency: Accuracy is contingent on the chosen local VLM (Gemini Nano, Gemma 3n, etc.). Performance may vary with dish complexity, video quality, and lighting.
     • Passive Income Limitation: As a free tool, it relies on Swiggy's ecosystem and builder programs rather than direct monetization, which may impact long-term development.
     • Support Scope: Order issues, refunds, and delivery problems are handled by Swiggy's support, not CraveLens, which may create a disjointed user experience for problems stemming from a misidentified item.

2. Key Alternatives & Differentiation:

   • Manual Swiggy/Zomato Search: The standard alternative. Differentiation: CraveLens automates visual search and cart building, removing manual text entry and comparison. It's intent-driven rather than query-driven.
   • Google Lens/ Pinterest Lens: General-purpose visual search tools. Differentiation: CraveLens is specialized for food-to-commerce. It doesn't just identify an item; it directly finds a purchasable, deliverable version from a specific service (Swiggy) and prepares the cart with user preferences.
   • Other Food Delivery Apps' In-App Camera Search: Some apps offer barcode or limited image search. Differentiation: CraveLens works passively on any content viewed in the browser, not just photos actively taken by the user. It's a discovery layer outside the delivery app, integrated into the content consumption experience.

Frequently Asked Questions (FAQ)

1. Is CraveLens safe? Does it spy on my browsing? CraveLens is designed with privacy-first architecture. Food detection uses on-device AI, meaning the analysis of video frames or selected images happens locally in your Chrome browser using your computer's resources. No video data, browsing history, or screen recordings are uploaded to external servers. Only after a food item is identified locally does the extension communicate with Swiggy's API to prepare a cart, sending only necessary text-based query data.

2. Does CraveLens automatically order food without my permission? No, CraveLens never automatically places an order. Its function ends at preparing a personalized Swiggy cart for your review. You must explicitly open the cart within the Swiggy interface, review the items, prices, and delivery details, and manually proceed through Swiggy's checkout process to confirm and pay for the order.

3. What food delivery services does CraveLens work with? Currently, CraveLens is exclusively integrated with Swiggy, India's major food delivery platform. This integration is facilitated through the Swiggy Builders Club program, which provides API access. It does not support Zomato, Uber Eats, or other international delivery services at this time.

4. What are the technical requirements to use CraveLens? You need the Google Chrome browser on a desktop or laptop. The extension requires WebGPU support (typically available in modern Chrome versions) to run the local AI models efficiently. A Swiggy account is required for the cart preparation and ordering functionality, though the food detection feature can be tested without one.

5. How accurate is the food recognition feature of CraveLens? Accuracy depends on the local Vision-Language Model (VLM) you configure (e.g., Gemini Nano, Gemma 3n) and the quality of the source image/video. It performs well on clear, well-framed dishes but may struggle with obscured foods, overly complex plates, or poor lighting. The extension displays a confidence percentage for each identification to set user expectations.