What’s Changed

Jev AI API adopts a different approach from traditional chat-type endpoints. This API is designed around three clear inputs: State, Model, Questions, and returns structured results for integrating AI judgments into application logic. Specifically, it supports three types of question types: Choice (classification/routing), Score (ranking evaluation), Noul (single Yes/No judgment), each corresponding to different use cases.

(Reference: Jev AI API Tutorial: Build Your First Structured Decision with Choice, Score, and Noul)

How It Works

The Jev AI API processing flow consists of the following steps:

  1. State Preparation: Provide context in text, JSON object, or array format.
  2. Question Type Selection:
    • Choice: Used for classification or routing. Example: Determining the department for a support ticket.
    • Score: Used for spectrum evaluation. Example: Calculating a risk score.
    • Noul: Used for a single Yes/No judgment. Example: Determining if a review is necessary.
  3. API Request Sending: Prepare an API key and send a minimal Noul request.
  4. Response Processing: Read the results for each question ID and handle probabilities as signals.

(Reference: How to Use the Jev AI Model: A Step-by-Step Developer Guide)

Migration Procedure

To integrate the Jev AI API into a server-side workflow, follow these steps:

  1. Verification in Jev AI Playground: Confirm the request structure and understand the result format.
  2. API Key Preparation: Obtain an API key from the Jev AI official website.
  3. Server-Side Code Integration: Send requests and reflect the returned structured results in application logic.
  4. Error Handling Implementation: Design fallback logic for timeouts or errors.

(Reference: Jev AI API Tutorial: Build Your First Structured Decision with Choice, Score, and Noul)

Performance Characteristics

Jev AI API has the following characteristics:

  • Non-Chat Design: Provides software-consumable structured judgments without replacing chat interfaces.
  • Support for Multiple Questions: Allows including multiple questions in a single request.
  • Probability Handling: Handles probabilities as judgment signals, considering the confidence of the final result.

(Reference: How to Use the Jev AI Model: A Step-by-Step Developer Guide)

Summary

  • Utilize Jev AI API to achieve structured judgments, such as prioritizing support tickets or calculating risk scores.
  • Prepare an API key, verify requests in Jev AI Playground, and then integrate into the server-side, incorporating AI judgments into application logic.
  • Leverage multiple question types (Choice, Score, Noul) and handle probabilities and confidence as signals to build flexible judgment logic.
  • Design error handling and timeout countermeasures in advance to build a reliable system.
  • Refer to Jev AI’s official documentation (Jev AI homepage) and consider the optimal implementation method for your company’s use case.