JevMade field notes / Technical guide

Use Jev as one step in an Effect-based AI assistant

effect-uai supplies small building blocks for custom AI assistants. It lets a program use Jev for focused decisions alongside other models that write text, while keeping the sequence of actions under program control.

Original by betalyraAgent workflows

Listen to this guide

JevMade’s plain-English explanation

0:00 /

AI narration

Credits and license

“effect-uai” by betalyra. Read the original source.

This expanded guide is an AI-narrated adaptation of the source’s essential explanation, examples and caveats, not a word-for-word reading. The synthetic voice does not imitate the author or imply their endorsement.

License: MIT

MIT License

Copyright (c) 2026 Betalyra Sociedade Unipessoal Lda.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.

Our summary

An AI assistant often repeats a cycle: inspect information, ask a model, and perhaps use a tool. effect-uai provides building blocks for that cycle rather than choosing the whole design. It uses Effect, a programming toolkit that makes failures and outside dependencies explicit.

The application keeps track of what has happened and decides which model or tool to call next. Its Jev connection supplies choices, ratings, and yes-or-no estimates. Other connections can write text, without pretending that a written reply and a focused decision are the same thing.

The project is experimental and warns that even smaller releases may break existing code. Building the loop leaves important work to its developers: limit repeated attempts, handle failed services, restrict tool permissions, and choose a stopping rule. This is not a ready-made assistant.

Key takeaways

Use Jev for a decision step, not as a writing model.
The application remains responsible for the order of actions.
Expect changes and test failure paths while the library is experimental.

More guides →See the experiments →