Should I use Jev for planning and scheduling?
Verdict for the sample idea
Run it on
a reasoning LLM
Every case needs generated text, planning or multi-step reasoning.
Sample idea
Given a list of 40 tasks with durations, dependencies and deadlines, and a team of six people with different skills, produce a weekly schedule that respects all constraints and tell us if the deadline is at risk.
Explanation
Two different problems live under this title, and Jev is the wrong tool for both.
Scheduling is constraint solving. There are durations, dependencies, capacities and deadlines, and one correct set of feasible answers. Code does this. Use a solver, or a topological sort with a calendar for the simple cases. The result is exact and explainable. "Is the deadline at risk?" is then a subtraction. Jev is weak at arithmetic, counting and date comparison, and a schedule is made of all three.
Planning is multi-step reasoning. Breaking "launch the feature" into ordered steps means that each step depends on the ones before it. Jev answers every question in one independent pass. It cannot think through step two after it decided step one. That is the work of a reasoning LLM, or of a person.
Jev can help with the inputs to both, where the judgment is local and the text is messy.
Which skill does this task need? That is a Choice over your six skill names, asked once per task description. Forty short calls turn free-text tickets into structured input for the solver.
How large is this task? That is a Score with levels written as situations: a one-line change, a change in one module, a change across several modules, a change that needs a migration. It is not an estimate in hours. It is a bucket that your code maps to hours from your own history.
Is this task blocked by something outside the team? That is a Noul.
The common trap is the agent loop. It is tempting to call Jev again and again, "what is the next step?", and to call the result a plan. Each call is a good local choice. Nothing checks the whole. Let the LLM or the solver own the plan, and let Jev make the fast judgments inside it.
Related tasks
Verdict produced by Jev on 2026-09-19, rubric version 2. Text written by a person. How the tool works: the report.