What happened
Alli AI shipped a WordPress plugin with a blunt mechanism: when an AI crawler requests a page, the plugin serves a complete, pre-rendered HTML version containing the full content; when a human visits, they get the normal site experience. It also deploys structured data automatically across the whole site, pitched as helping both traditional and AI search engines extract and cite content accurately. Pricing wasn't disclosed in Practical Ecommerce's July 7 roundup.
The problem underneath is legitimate. Plenty of AI crawlers execute little or no JavaScript, so a page that assembles its content client-side can read as half-empty to the systems deciding what to cite and recommend. Pre-rendering genuinely fixes that. The question the plugin answers badly is whether you fix it for one audience or for everyone.
Why this matters
Serving crawlers different HTML than humans is cloaking-adjacent territory. The intent here isn't deceptive — same content, different rendering path — but you're now maintaining two versions of every page and betting they never drift apart. They will drift. A price updates in one and not the other, a product change misses the pre-render cache, and suddenly the AI systems you were optimizing for are quoting stale data with your name attached. The failure mode isn't a penalty; it's an AI assistant confidently telling a shopper your old price.
Google's own team keeps cautioning against building parallel AI versions of a site, and the durable engineering pattern agrees with them: one page, rendered as real HTML for every reader, human or machine. Server-side rendering solves the crawler problem without the fork, and there's nothing to keep in sync afterward. A user-agent-switched copy is the kind of shortcut that feels clever for exactly as long as nobody has to maintain it.
If you're on BigCommerce, note that you mostly don't have the disease this treats. Stencil storefronts ship server-rendered HTML, so product names, prices, and copy sit in the page source without JavaScript running. Bolt-ons like this exist because some platforms made client-side rendering the default; check whether yours did before buying the cure.
What to do about it
Check what AI crawlers actually get from your pages
Fetch your top product and category pages with JavaScript disabled, or just view source and search for the price and key specs. If the facts are present in raw HTML, this entire tool category has nothing to sell you. Ten minutes of checking beats a subscription bought on fear.
If content is missing without JS, fix rendering once
Move the affected content to server-side rendering or a pre-render served to every visitor, not a bot-only fork. One version means nothing to sync, nothing to drift, and no awkward conversation about why crawlers see a different page than customers do.
Audit any auto-deployed schema before trusting it
Site-wide automatic structured data ships wrong values just as confidently as right ones. If a tool generates your markup, validate a sample of pages by hand and confirm the schema matches what the page displays. A mismatch between markup and page hurts more than missing markup ever did.