# AgentSite vs Prerender.io — an honest comparison

Prerender.io has been the JS-rendering incumbent since 2013. AgentSite is a younger product taking a different shape. Here is what each one does, and which one to pick.

By AgentSite · Updated 2026-06-25

## What is [Prerender.io](http://Prerender.io)?

[Prerender.io](http://Prerender.io) launched around 2013 and is the longest-running JS-rendering service for SPA SEO. The product renders a customer's site in headless Chromium and serves the cached HTML to bots that don't execute JavaScript. Their pivot to AEO/AI-search messaging is recent but explicit: the homepage now reads _"Get your site discovered everywhere search happens"_ and _"AI crawlers won't execute JavaScript. Without a rendering solution like [Prerender.io](http://Prerender.io), your site's important information may not show up in AI search results at all."_ ([prerender.io](https://prerender.io)). Named customers include Walmart, Spotify, and Netlify; the company is privately held with no disclosed funding round on Crunchbase.

The product is delivered via DNS, middleware, or a CDN integration, and renders are cached globally. Pricing is Free → deprecated as of October 2025 (1,000 renders/mo), Starter $49/mo (25K renders), Growth $149/mo (100K renders), Pro $349/mo (500K renders, with on-demand cache invalidation), and Enterprise Plus on request (from 5M renders). All plans include unlimited domains. On-demand cache invalidation requires the $349/mo tier.

The output is crawler-readable HTML. [Prerender.io](http://Prerender.io) does not generate llms.txt, per-page markdown mirrors, JSON-LD schema, well-known endpoints, or any AI-bot traffic dashboard. Customers who want those layers typically stack a separate AEO tool on top.

## Feature-by-feature, observed

| Capability | [Prerender.io](http://Prerender.io) | AgentSite |
| --- | --- | --- |
| JS rendering for SPAs | Yes — the incumbent | Yes |
| Output format | Crawler-readable HTML | HTML + per-page markdown mirror + JSON-LD + llms.txt + well-known endpoints |
| llms.txt generation | No | Yes |
| AI bot traffic dashboard | No | Yes (via install snippet) |
| Same DOM for humans and bots | No — UA-branched rendering | Yes — same bytes served to all visitors |
| Free tier | Deprecated October 2025 | Yes — first Agent Experience Report free, no signup |
| On-demand cache invalidation | $349/mo tier and above | All paid tiers |
| Install patterns | DNS / middleware / CDN integration | Nginx, Express, Express-Sidecar, Edge, streaming-SSR SDK |
| Enterprise procurement (SOC 2, named accounts) | Yes — Walmart, Spotify, Netlify | Not pursued |
| Time in market | ~13 years (founded 2013) | Newer entrant |
| Entry price (after free deprecation) | $49/mo (25K renders) | $29/mo |

## The philosophy difference

[Prerender.io](http://Prerender.io) was built for the SEO era and earned its incumbency by solving one specific problem extremely well: serve crawler-readable HTML to bots that don't run JavaScript. The architecture is UA-detection plus a global cache. That focus is what gets you to Walmart and Spotify — a single-purpose tool with thirteen years of reliability data, enterprise procurement, and a CDN that handles billions of requests. The cost of that scope is that the output is _only_ HTML; everything AEO-specific (markdown for LLMs, llms.txt, schema, agent-card files, bot traffic telemetry) sits outside the product. Customers who want those layers compose them themselves or buy a second tool. The October 2025 free-tier deprecation also closed off the no-signup evaluation path, which raises the friction for indie users to try the product.

AgentSite arrived after Prerender's foundation was already poured, and we are not trying to out-incumbent them on raw render volume. The bet is to ship the whole agent-experience stack as one install. A single render pass produces crawler HTML _and_ per-page markdown, JSON-LD, llms.txt, and well-known endpoints. The same snippet that injects them also counts bot UA traffic and feeds the Agent Experience Report. The architectural commitment is "same DOM for everyone," trading one optimization angle for a different posture against UA-branching policy risk. The buyer is a developer who wants the whole stack handled, not a procurement office buying a render service.

## When to pick [Prerender.io](http://Prerender.io), when to pick AgentSite

**Pick [Prerender.io](http://Prerender.io) if:**

-   Your only problem is "AI crawlers don't run JavaScript" — you don't want llms.txt, schema injection, or scoring
-   You need a vendor with 13 years of enterprise track record (Walmart / Spotify / Netlify lineage)
-   You need 500K+ renders/mo at $349/mo and can pay for on-demand cache invalidation
-   Your procurement office requires a vendor it can already trust
-   You're comfortable that the output is HTML-only and you'll compose other AEO tooling separately

**Pick AgentSite if:**

-   You want one install that handles rendering + markdown + schema + llms.txt + well-known + scoring
-   You want bots and humans to see the same DOM (no cloaking-risk posture)
-   You want a free first report with no signup before you decide to pay
-   You're an indie SaaS founder or growth engineer, not a procurement office
-   You want the AI bot traffic dashboard and Agent Experience Score built in

See [vs Encited](/agent/vs-encited) or [vs DataJelly](/agent/vs-datajelly) for the newer-generation prerender alternatives, or [run a free Agent Experience Report](/score).