Comparisons · Last updated 2026-09-21

Jev vs email automation

Email automation (Braze, Customer.io, your ESP) owns templates, cadence, and deliverability. Jev can decide whether this user/this event should enter a send path.

Independent. Jev does not generate the email (jaggedness: generation). docs.typesafe.ai. No keys.

Comparison scope

“Has this ticket already been promised a refund?” can be a Noul on CRM notes before a win-back send. “Send Tuesday 10:00 in the user’s TZ” is the ESP.

Criteria that decide the architecture

Axis	Jev (System One)	Email automation
Deliverability	None	Wins
Template merge	None	ESP
Should we send?	Noul + confidence	Segment rules
Write the body	LLM / human	Template

Decision quality and control

A wrong send is a trust event. Fail closed on timeout. Do not let a 0.51 Noul fire a discount blast.

Integration trade-offs

ESP webhook → build small state (last tickets, last send id) → Jev → skip/send/review. Keep unsubscribe and legal suppression out of the model — those are rules.

TypeSafe’s public models page lists jev-1.13 at $0.042 per million input tokens with output tokens free — a vendor claim, not a jev.pro measurement. Confirm on the models page before you bid.

When each approach fits

Prefer Jev when

Send/no-send depends on messy notes

Prefer Email automation when

Cadence, ESP, and compliance suppression

Rules suppress; Jev judges residual; ESP sends; LLM optional for copy.

What this page does not claim

Not an ESP review.
No open-rate claims.
Schema-safe output is not the same as factual correctness.
No independent bake-off numbers live here.

When not to use Jev

Do not use Jev to deliver mail, manage suppression, or honor CAN-SPAM / consent. Those are the ESP and your code. A schema-safe send is not consent.

Do not ask Jev to write the paragraph. Jaggedness: generation. It may pick a template Choice. An LLM or a human drafts; the ESP merges.

Do not fire a blast on a 0.51 Noul. Fail closed on timeout. Wrong send is a trust event, not a retry.

Skip Jev when the segment is already field-complete and there are no messy notes to judge.

Pairing-specific failure modes

Quiet hours as language. “Is it morning in the user’s TZ?” is the ESP’s scheduler, not a Noul.
Last-send identity. “Did we already promise a refund?” needs the ticket span in state — Jev cannot browse the ESP.
Prompt-only copy. A generator with no decide step is prompt-only. Decide, then draft.
Alias on the gate. Pin jev-1.13.0 after you fit send/skip/review floors; jev-latest changes who gets the win-back without a segment edit.

Production sketch

Suppression, unsubscribe, and legal lists run in code/ESP first. Those users never reach Jev.
Webhook builds small state: last tickets, last send id — not the full CRM novel.
POST send / skip / review with confidence. Pin jev-1.13.0. Fail closed on timeout or 429.
send + high confidence → ESP template (optional LLM draft after the decision). review → human. skip → no mail.
Log response.model, template id, and whether the ESP accepted. CRM notes stay on CRM workflows. No open-rate claims here.

Hub: Comparisons. Siblings: vs crm workflows, vs prompt only systems.

FAQ

Can Jev personalize copy? It can pick a template Choice. It should not write the paragraph.

CAN-SPAM / consent? Code and your ESP. Not a Jev question.

Decide-then-draft? Yes. Jev gates; LLM/human drafts; ESP sends. Do not invert that.

Disclaimer

This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai. Never treat jev.pro as TypeSafe official documentation. We do not sell, issue, or proxy API keys.