JevClassifier

JevClassifier and JevClient answer classifier questions through Jev, TypeSafe’s hosted classification model, with calibrated probabilities.

Overview

JevClassifier answers classifier questions through Jev, TypeSafe’s hosted classification model. All the questions about one state go to Jev in one request, and an answer typically comes back in about a tenth of a second. Jev’s probabilities are calibrated, so a threshold such as “act when the probability is above 0.8” means the same thing from one question to the next.

A JevClassifier asks through a JevClient, which holds the HTTP/2 connection, adds the auth header, retries when Jev is busy, and counts tokens. Build the classifier with an API key to give it a client of its own, or pass a JevClient to share one connection between several classifiers.

Installation

uv add "pipecat-ai[jev]"

Prerequisites

A Jev API key from TypeSafe, usually set as an environment variable:

TYPESAFE_API_KEY=...

Configuration

JevClassifier

from pipecat.classifiers.jev.classifier import JevClassifier

api_key · str | None · default: "None"

Jev API key, when the classifier should have a client of its own. One of api_key and client is required.

client · JevClient | None · default: "None"

A client to share with other classifiers. One of api_key and client is required.

name · str | None · default: "None"

Name of the classifier, as it appears in logs and metrics.

JevClient

from pipecat.classifiers.jev.client import JevClient

api_key · str · required

Jev API key.

base_url · str · default: "https://api.typesafe.ai"

Where the API is served.

model · str · default: "jev-1.13.0"

The Jev model to ask. It is pinned so thresholds you tune against it keep holding. jev-latest follows TypeSafe’s newest release.

timeout · float · default: "10.0"

Seconds to wait for a reply before raising ClassifierError.

max_retries · int · default: "3"

How many times to retry a request Jev refused because it was busy (HTTP 429 or 529), with exponential backoff.

Usage

Basic Usage

import os

from pipecat.classifiers.base_classifier import YesNoQuestion
from pipecat.classifiers.jev.classifier import JevClassifier

classifier = JevClassifier(api_key=os.getenv("TYPESAFE_API_KEY"))

results = await classifier.yes_no(
    "Hi, you've reached Dana. Leave a message.",
    {"voicemail": YesNoQuestion(instructions="is this a voicemail greeting?")},
)
results["voicemail"].is_yes  # True
results["voicemail"].probability  # 0.97

Most of the time you don’t call the classifier yourself: you pass it to a component that asks it, such as VoicemailDetector or UIWorker.

Sharing a Client

Several classifiers can share one JevClient, and with it one connection pool and one token count:

from pipecat.classifiers.jev.classifier import JevClassifier
from pipecat.classifiers.jev.client import JevClient

client = JevClient(api_key=os.getenv("TYPESAFE_API_KEY"))

voicemail = VoicemailDetector(classifier=JevClassifier(client=client))
ui_worker = MyUIWorker("ui", llm=ui_llm, classifier=JevClassifier(client=client))

A client the classifier created is closed in the classifier’s cleanup(). A shared client is left open, so close it yourself with await client.close() when every classifier using it is done.

Opening the Connection Early

setup() opens the connection to Jev ahead of the first question, so the first answer does not pay for the TLS handshake. Components that own a classifier, such as VoicemailDetector, call it for you. If the connection cannot be opened at setup, a warning is logged and the first question opens it instead.

JevClassifier Properties

Property	Type	Description
client	JevClient	The client this classifier asks through.
model	str	The Jev model the questions go to.

JevClient Reference

Properties

Property	Type	Description
model	str	The Jev model the questions go to.
usage	JevUsage	Tokens used so far over every request, as input_tokens and output_tokens.

Methods

connect

await client.connect()

Opens the connection to Jev. It connects once: a client shared by several classifiers is connected by each of them, and only the first call sends anything. Raises ClassifierError if Jev could not be reached or refused the request.

close

await client.close()

Closes the connection pool.

Metrics

After every call, a JevClassifier reports the time it took as ProcessingMetricsData and the tokens Jev used as LLMUsageMetricsData through its on_metrics event.

Notes

• Limits: a choice question takes at most 255 options, available as JEV_MAX_CHOICE_OPTIONS in pipecat.classifiers.jev.classifier. A question with more raises ClassifierError before anything is sent.
• Errors: a rejected request, a busy Jev after every retry, an unreachable server, or a reply missing an answer all raise ClassifierError.
• Idle connections are kept open for 240 seconds, so a gap between questions does not cost a new TLS handshake.