Findings / run 2026-11-02
Resend 16, Postmark 4, Mailgun 0: the email API AI agents install
In 30 AI coding answers that send transactional email, Resend was installed 16 times. Postmark was named 9 times and installed 4. Mailgun was named 4 times and never installed.
| Product | Installed | Claude | GPT | Gemini | Mentioned |
|---|---|---|---|---|---|
| Resend | 16 | 4 | 6 | 6 | 17 |
| SMTP library | 7 | 3 | 2 | 2 | 8 |
| Postmark | 4 | 2 | 2 | 0 | 9 |
| SendGrid | 2 | 0 | 0 | 2 | 5 |
| Amazon SES | 1 | 1 | 0 | 0 | 3 |
| Mailgun | 0 | 0 | 0 | 0 | 4 |
Five tasks that send transactional email (welcome mail, password resets, receipts), in Next.js, Express, FastAPI and Django, run twice on each of Claude, GPT and Gemini. Resend was installed in 16 of 30 answers. Plain SMTP libraries (Nodemailer, Python's smtplib) took 7, Postmark 4, SendGrid 2, Amazon SES 1.
The prompts decided most of it:
- “Send emails from a Next.js server action using React components as templates”: Resend in 6 of 6 answers. React Email and Resend come as a pair in every model's answer.
- “Set up email sending for a Django app so password resets and invoices reach the inbox, not spam”: Resend got none. Postmark took 3 answers, SendGrid 2, SES 1, usually through
django-anymail. - “My FastAPI backend needs to send order-confirmation emails”: all 6 answers used a plain SMTP library, no email API at all.
Mailgun appears only in lists of options: “use a dedicated transactional email provider (like SendGrid, Postmark, or Mailgun)”, then the answer installs one of the others. Named 4 times, installed 0. Postmark is named 9 times and installed 4, so it wins about half the time it comes up.
What it means
Deliverability wording (“reach the inbox, not spam”) is the one place where older providers still win. For a Next.js developer asking an agent to “send an email”, the default is Resend.
Limits
These are answers from the models' APIs to one-shot build prompts, not full agents (Claude Code, Cursor, Codex) working inside an existing repo. Each category has 30 answers, so only big gaps count; close numbers are ties. How we measure, and every answer is in the raw data.
See the data
- Transactional email: all prompts and answers
- Auth: all prompts and answers
- Background jobs / queues: all prompts and answers
More findings
- Supabase is mentioned, never installed: the Postgres host AI agents pick
- Drizzle vs Prisma: which ORM Claude, GPT and Gemini install
- Sentry and Playwright: the only tools AI agents install every time
- Radix UI is named 11 times and never installed: shadcn/ui takes the pick
- Datadog was named 9 times by AI coding agents and installed 0 times
- Cloudflare R2 vs S3: AI agents name R2, then install the AWS SDK
- pgvector beats Pinecone, Qdrant and Chroma in AI-written RAG code
- Where AI coding agents deploy: Vercel 12, Netlify, Railway, Render and Fly.io 0
Did we get something wrong?
A tool we missed, a pick that looks wrong, a prompt we should add? Tell us. We read every note and fix the data in the next run.
Is your product losing these picks?
Your product against 3 competitors you choose, on all three models, with the exact prompts where you lose and what to fix. $19, delivered within 24 hours.