Comparisons · Last updated 2026-09-21

Jev vs logistics dispatch

Logistics dispatch solves assignment and routing under constraints. Jev does not minimize miles. It can label a driver or customer exception note.

Independent. Official: keep numeric optimization in code. docs.typesafe.ai.

Comparison scope

ETAs, capacity, and VRP stay solvers. “Customer says gate code changed and dock is blocked” is a Choice over exception types.

Criteria that decide the architecture

Decision quality and control

A wrong exception class delays a truck. Use unknown and a human dispatcher on low confidence. Do not ask Jev which stop is closer.

Integration trade-offs

TMS exception webhook → note text → Jev → reason code → solver or human. Pin the model.

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

• Free-text exceptions at volume

Prefer Logistics dispatch when

• The hard problem is the route

Solver dispatches; Jev labels stories; dispatchers handle unknown.

Dispatcher still owns the radio

A wrong exception class is recoverable if unknown exists. A Jev-picked driver assignment is an optimization fantasy. Keep VRP in the solver.

What this page does not claim

• Not a TMS replacement.

• Schema-safe output is not the same as factual correctness.

• No independent bake-off numbers live here.

When not to use Jev

Do not ask Jev which truck is closer, which window fits, or who has remaining hours. Those are solver constraints and arithmetic. Jaggedness: math and dates are text, not quantities — compare miles and appointment times in code.

Do not feed POD photos as if Jev were a vision model. OCR first; then a Choice on the extracted note.

Do not let a schema-safe exception class auto-reoptimize the route without a dispatcher when confidence is low. A well-typed dock_blocked can still be a weather delay.

Skip Jev when the exception is already a structured TMS code, or when the payload is only lat/long and time windows.

Pairing-specific failure modes

• Address-as-language. “Drop at the second gate after the rail spur” is not a geocode. Geocode in code; ask Jev only for the exception type.
• Multi-stop indirection. “Is the parent DC of this SKU’s issuer on the same run?” is a graph walk. Official jaggedness: reduce hops.
• Alias move on a live yard. Floors fitted on jev-1.13.0 drift if the TMS worker ships jev-latest.
• Radio bypass. Auto-texting the driver from a Jev label without unknown is how a wrong class becomes a missed appointment.

Production sketch

1. TMS webhook: exception note + already-structured stop ids (no full route table in state).
2. OCR any photo to text. Regex appointment stamps; compare them in code.
3. POST a Choice over your exception taxonomy plus unknown. Pin jev-1.13.0. Log response.model.
4. High-confidence mapped codes may open a solver re-opt or a playbook. unknown or low confidence → dispatcher owns the radio.
5. Never write a new driver assignment from Jev. The solver proposes; a policy Noul (“needs hazmat-certified”) may filter the proposal.

Hub: Comparisons. Siblings: vs orchestration platforms, for operations.

FAQ

Can Jev pick the driver? Not as an optimization. Maybe a policy Noul (“needs hazmat-certified”) after the solver proposes.

Addresses? Geocode in code.

Is this a VRP product compare? No. Cover the intent “language exception vs solver,” not a vendor bake-off.

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.