Jev versus churn models: a practical comparison
A churn model is usually a trained classifier or survival model on product events. Jev does not replace that table. It can turn support threads and NPS comments into extra features.
Unofficial. TypeSafe’s AutoResearch cookbook even treats Jev outputs as features for a classical model — composition, not “Jev is churn AI.” docs.typesafe.ai.
Comparison scope
Login decay, invoice fails, and seat count stay in the warehouse. “Sounds like they are leaving” is a Noul on the latest tickets.
Criteria that decide the architecture
| Axis | Jev (System One) | Churn models |
|---|---|---|
| Primary data | Text you send this call | Event store |
| Calibration to 90-day churn | You must measure | The model’s job if trained |
| New comment today | Immediate | Needs a pipeline |
| Numeric tenure | Code / warehouse | Feature |
Decision quality and control
A Noul is not P(churn in 90 days) unless you calibrated it that way on your labels. Do not present it as a survival curve.
Integration trade-offs
Nightly: event-model score. On ticket: Jev language Noul/Score. Combine in code. CS queue on OR of high event-risk and high language-risk.
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
- Qualitative signals arrive as text and you need them the same day
Prefer Churn models when
- You have a maintained event model and little text
Warehouse model + Jev text features + playbook. Humans still own win-back offers.
What this page does not claim
-
No AUC vs your data-science model.
-
Schema-safe output is not the same as factual correctness.
-
No independent bake-off numbers live here.
FAQ
Can Jev predict a date of churn? Dates are jagged. Keep timelines in code.
Fine-tune on churn labels? Not per TypeSafe: no customer LoRA. Use outputs as features instead.
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.
Hub: Comparisons. Siblings: vs lead scoring, usecase churn decision workflow. Canonical: https://docs.typesafe.ai.
Sources
Public TypeSafe or adjacent documentation only. No private claims.