
Google Cloud Vision
A customer sends a photo of a product and asks if you carry it. Your AI agent runs it through Google Cloud Vision, identifies the item, matches it against your product catalog, and responds with pricing and availability. Image recognition that actually serves customers.




Your AI agent uses Google Cloud Vision to interpret images customers share, from product photos to document scans, turning visual data into actionable responses.
Google Cloud Vision
Product identification, visual search, and image-based support become possible when your AI agent can see what your customers see.
A customer photographs a lamp they saw at a friend's house and sends the image to your chat. Your AI Agent runs it through Google Cloud Vision Product Search, matches it against your home goods catalog, and returns the exact product page with price and stock status. The customer adds it to their cart without ever typing a product name. Visual discovery that converts browsers into buyers.
A field technician photographs a worn-out component and messages your support chat. Your AI Agent sends the image to Google Cloud Vision, identifies the part from your reference image library, and provides the part number, compatibility list, and ordering link. The technician orders the replacement in minutes instead of flipping through a 500-page parts manual.
Your merchandising team needs to add 200 new products to the visual search index before a seasonal launch. Your AI Agent batch-imports product images into Google Cloud Vision, creates Product Sets for the new collection, and monitors the async operation until indexing completes. The entire catalog goes live for visual search without manual uploads to the Vision API console.

Google Cloud Vision
FAQs
The agent sends the customer's image to Google Cloud Vision's Product Search API along with your Product Set ID. Vision AI compares visual features against all reference images in your catalog and returns a ranked list of matches with confidence scores. The agent then presents the top results with product details.
Google Cloud Vision Product Search supports homegoods, apparel, toys, packaged goods, and a general category. Each product type has optimized visual matching algorithms. If your products fall outside these categories, the general category provides broad visual similarity matching.
You can use the Tars agent to create products, add reference images from multiple angles, and organize them into Product Sets. For bulk onboarding, the agent supports CSV-based batch imports through Vision API's async import endpoint. Start with 3-5 reference images per product for best accuracy.
Tars passes customer images to Google Cloud Vision for analysis during the conversation. Images are not permanently stored in our systems. Your Google Cloud project's data retention and processing policies govern how Vision API handles the image data on Google's infrastructure.
Tars requires a Google Cloud Vision API key with access to the Product Search and Vision AI endpoints. The API key should be scoped to your specific Google Cloud project. You manage billing, quotas, and access controls through your Google Cloud Console.
Yes. When deployed on WhatsApp, web chat, or other channels, the agent receives the image attachment and forwards it to Google Cloud Vision for analysis. The visual search works identically regardless of which messaging platform the customer uses.
Accuracy depends on the quality and quantity of reference images in your catalog. Products with 5-10 reference images from different angles and lighting conditions match significantly better than single-image entries. Confidence scores are returned with each match, so the agent can set thresholds for when to show results versus ask for a better photo.
Reverse image search scans the entire internet. Vision API Product Search matches only against your curated product catalog, returning items you actually sell with your pricing and availability data. It is purpose-built for retail and product identification, not generic web image matching.
Don't limit your AI Agent to basic conversations. Watch how to configure and add powerful tools making your agent smarter and more functional.

Privacy & Security
At Tars, we take privacy and security very seriously. We are compliant with GDPR, ISO, SOC 2, and HIPAA.