Exa 3.0 indexes 1.27B product listings for AI agents
Exa 3.0 adds 1.27 billion ecommerce listings with prices and availability to its AI agent search API. What that means for your product pages and your feed.
AI shopping agents do not browse your homepage. They query a search API, and the answer is whatever that index says about you. Exa 3.0, the new release of the search API that many AI agents use, now carries 1.27 billion ecommerce listings with prices and availability, according to Exa's announcement. If your product sells online, there is a good chance an agent now quotes your price from Exa's copy of your page, not from your checkout.
What actually happened
Exa announced Exa 3.0 on October 19. The key parts:
- New data in the index: 1.27B ecommerce listings with prices and availability, 211M+ local businesses with hours and ratings, plus news, sports, weather, flights and events.
- Its own crawler: Exa says it tracks more than 2 trillion URLs and serves 100 billion pages from an independent index.
- Two price tiers: Exa Lite at $4 per 1,000 requests for cost-sensitive apps, and Exa Auto at $7 per 1,000 with more compute on ranking. Both run on the same index.
- Dynamic Highlights, now GA: Exa says it returns the relevant passages across results with 95% fewer tokens than full page text, at no extra charge.
Exa also says it ranks first on eight benchmarks, including 91.8% on BrowseComp. Exa ran those tests itself, so treat them as vendor claims. The announcement does not say how often listings refresh or how it extracts price and stock. Those are the two details that matter most to a merchant.
Why it matters for your business
Your page is the feed now. Google Merchant Center gets a feed you control. A crawler-built index gets whatever your product page shows on the day it visits. If the price is buried in JavaScript or the stock badge says "In stock" on a sold-out variant, that is what an agent repeats.
Stale data costs you twice. An agent that quotes a lower price than your checkout loses the sale and the trust. One that thinks you are out of stock never sends the buyer at all.
Make it easy to read. Put the price and availability in server-rendered HTML, add Product structured data with offers.price and offers.availability, and keep variant URLs clean. You cannot control Exa's refresh cycle. You can control what it finds when it arrives.
For builders, the price is right. At $4 per 1,000 searches, product lookup inside your own agent or support bot costs fractions of a cent per call.
Key takeaways
- Exa 3.0 adds 1.27 billion ecommerce listings with prices and availability for AI agents
- Exa Lite costs $4 per 1,000 requests; Exa Auto costs $7
- Benchmark wins are self-reported, and refresh frequency is not disclosed
- Agents may quote your price from a crawled copy, so your product page must be accurate and readable
- Server-rendered prices and Product structured data are the cheapest fix
What does an AI agent see when it reads your catalog? We audit your product pages for agent readability, fix structured data and rendering, and build agent search into your own tools. Ask us to run an agent-readiness check on your store, or see what we build for commerce teams.
Sources: Exa: Introducing Exa 3.0.
- #ai agents
- #agentic commerce
- #search
- #product data
- #exa
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Visa, BILL, Billtrust Pilot: AI Agents Pay Supplier Invoices
Visa, BILL and Billtrust run a live pilot where an AI agent pays approved invoices through supplier portals with Trusted Agent Protocol. No shared logins.
Read itShopify UCP location search: AI agents can find your stores
Shopify's UCP 2026-08-25 lets AI shopping agents search a merchant's retail stores by city, distance and item. Your location data is now agent-facing.
Read it