# Resend pricing teardown · 90/100 · Onto > Resend is the rare SaaS company that already speaks Markdown to AI agents. Native content negotiation, JSON-LD schema present, robots open, 0.9% React Tax. Every check passes. The remaining 10 points are about observability and policy — not infrastructure. This is what agent-ready actually looks like. **Source:** /case-studies/resend-pricing **Extracted:** 2026-05-20T20:59:17.658Z --- TeardownResend · May 2026 ## Resend already serves AI-ready Markdown. _Here's what 90/100 looks like._ Resend is the rare SaaS company that already speaks Markdown to AI agents. Native content negotiation. JSON-LD schema present. Robots open. **0.9% React Tax**. Every one of the five agent-readiness checks passes. The remaining 10 points are about observability and policy — not infrastructure. This is the closest thing we've published to a reference implementation. [Live target · resend.com/pricing](https://resend.com/pricing) 90/100 Agent-readyAIO Score · the highest we've published Read time5 min read ScannedMay 2026 // TL;DR Resend made the architectural decisions most SaaS companies still haven't: server-rendered HTML, low JS overhead, JSON-LD on every product surface, and — uniquely — an Accept-header negotiation path that serves Markdown to AI clients. None of this is hard. All of it is rare. The next 10 points come from layering analytics and policy on top — seeing which agents read which routes, when, and with what effect. 01 // Pillar breakdown ### All three pillars in the green. The AIO score is subtractive — penalties trace to specific structural issues. Resend has no penalties at the pillar level. The 10-point gap is in the soft layer above: observability, policy, and analytics. 01Weight · 40% React Tax How much of the HTML is noise? 0.9%of bytes are visual noise Less than 1% bloat — astonishing for a modern SaaS site. Resend ships server-rendered HTML, minimal client JS, and structures content semantically. This is what a healthy code-to-content ratio looks like. Raw payload 453 KB Semantic text 12,441 chars 02Weight · 35% Semantic Richness Does the HTML tell a structured story? 2 / 2structural checks pass Strong heading hierarchy, JSON-LD schema present, proper landmark elements. Resend treats Schema.org markup as a first-class concern, not an afterthought. AI agents can verify product, pricing, and offer facts directly from the markup. JSON-LD schema Present Heading hierarchy Strong 03Weight · 25% Content Negotiation Does the server speak Markdown? Supportednative Accept-header negotiation The rare site that already routes AI agents to Markdown surfaces. Send Accept: text/markdown — Resend returns Markdown. This is the application-layer move that 99% of SaaS sites still haven't made. Resend already shipped it. Markdown serving Yes Robots.txt Allowed 02 // Agent-readiness checklist ### Five of five. This is the bar. Every check passes. The infrastructure side, the application side, the protocol side — all green. We can count the SaaS companies that hit this on one hand. 01Robots.txt allows agent crawlingPass 02WAF passing agent trafficPass 03Native Markdown content negotiationPass 04JSON-LD schema verificationPass 05Deep semantic hierarchy (H1–H3)Pass 03 // What an AI agent receives ### Rich, structured, low hallucination risk. Resend's clean output is dense — pricing tiers, add-ons, FAQs, pay-as-you-go pricing, all preserved with their heading hierarchy intact. `1.0 KB` of usable content with zero ambiguity for the model. .onto/md · resend-pricing · 1.0 KB \# Pricing · Resend \> Start for free and scale up to millions of emails with flexible plans. \*\*Source:\*\* https://resend.com/pricing \## Pricing \### Add-ons \### Dedicated IPs - $30 / mo \### Sending & receiving \### Deliverability & reliability \### Security & privacy \### Customer support \### AI credits \### Frequently asked questions \#### Do you offer any discounts for annual subscriptions? \#### Do you offer non-profit or education pricing? \#### What payment methods do you accept? \#### Is there a free trial available? \#### Who can I contact about a custom plan? \#### What happens if I send more emails than my plan allows? \#### What is the difference between transactional and marketing emails? \### Pay-as-you-go \#### Emails \#### Automations · $0.0015 / per run \# Protocol · Semantic\_Extraction\_v1.0 \# Hallucination risk · Low 446×Smaller 115,968 → 260Tokens (raw → clean) LowHallucination risk 04 // The founder asked us directly ### How Resend gets to 100/100. When we published this teardown, the founder of Resend replied publicly. The remaining 10 points aren't about fixing infrastructure — they're about layering observability and policy on top of what Resend already ships. Public reply on X · March 20, 2026 When this teardown went live, Resend's CEO replied to us publicly — asking the exact question that frames the rest of this post. [ ZR Zeno Rocha@zenorocha · CEO, Resend how can we get to 100/100? March 20, 2026View on X ](https://twitter.com/zenorocha/status/2034813547897589842) 05 // Our answer ### Only one recommendation — and it isn't infrastructure. There's nothing structural left to fix on the Resend pricing page. The remaining points come from making the existing agent surface observable and governable. 01Medium priority Migrate to Onto SDK Native Markdown is shipping — what's missing is observability and policy. The Onto layer adds bot analytics, per-route context injection, and rate-limit controls without changing the negotiation behaviour already in place. [Talk to us](/serve-sdk) 06 // What Onto still adds ### Latency, cost, and analytics — without changing what Resend already ships. 446× Payload reduction 453 KB → 1.0 KB 100% Token waste eliminated 115,968 → 260 tokens $0.0026 Cost / GPT-4 query was $1.160 · with Onto <15 ms Edge response p99 · CDN edge node The reference implementation ### Build like Resend. Measure with Onto. Run the same five checks against your own URL. If you score above 80, we want to talk to you about what comes next. [Scan your site](/#scanner)// No card · No signup · 4 seconds [← All teardowns](/case-studies)[Linear pricing teardown →](/case-studies/linear-pricing)