Hi everyone!
I’ve released the first test version of Jev for Homey, powered by TypeSafe AI.
Jev helps your Flows make decisions based on the context you provide. Which lighting scene fits the evening? Can a notification wait? How much extra light would be useful right now?
You describe the situation using text and Flow tags, then ask a question directly on a Flow card. You decide which information matters. The app doesn’t scan your devices or automatically collect household data.
What can it do?
All four cards are available under Then:
- Ask a yes/no question — returns a boolean answer and the probability of “yes”.
- Choose from a list — provide possible answers, one per line, and get the selected answer and its number back.
- Score a situation — describe three levels and receive a score from 0 to 2, including values in between.
- Evaluate JSON — an advanced option for supplying structured context and questions.
Use the result tags in Advanced Flow to decide what happens next. The simple cards have adjustable acceptance thresholds. If Jev is uncertain, the card follows the error path so you can handle that in your Flow.
A simple example
State:
We’re watching a film, we have guests, and it’s dark outside.
Question:
Which lighting scene would fit best?
Answers:
Film: dim lighting for watching television.
Cozy: enough warm light to talk with guests.
Bright: enough light for reading or working.
Connect the result to your existing lighting actions. The context, options and actions are all yours to define.
Getting started
You’ll need a local Homey running 12.4 or newer, a TypeSafe API key with credit, and Advanced Flow to use the result tags in subsequent cards. Add your key in the app settings, then configure everything else on the Flow cards.
The context and questions you supply are sent to TypeSafe for evaluation. API calls, including connection tests, use your TypeSafe credit.
