---
title: "Release Notes: v3.4"
description: "Voicetta v3.4 release notes covering AI Evaluations for automatic call grading, evaluation results in History, and centrally managed analysis models."
publishedAt: "2026-07-28"
modifiedAt: "2026-07-28"
category: "Release Notes"
tags: "Voicetta, v3.4, release notes, AI evaluations, call grading, quality assurance"
navOrder: "97"
canonical: "https://voicetta.com/docs/release-v3-4"
---

# Release Notes: v3.4

## Why it matters for your business

- **Standardize service:** define the behaviors your agent must show on every call and check them automatically.
- **Work faster:** no more listening to random call samples — every call is graded the moment it ends.
- **Save money:** spot failing behaviors early, before they cost bookings or repeat Guests.
- **Delight every Guest:** measurable call quality means problems are fixed based on data, not guesswork.

## Short Version

Version `v3.4` adds **AI Evaluations** — automatic grading of every call against your own quality criteria — with results in call details and aggregate pass rates on a new AI Evaluations page.

## What Was Added

### AI Evaluations

You can now define evaluation criteria for your agent — a short title plus a description of the expected behavior. For example:

- **Stay booked:** each time the caller wanted to book a stay, the agent booked it.
- **Callback offered:** when the agent could not help, it offered a callback.

After every call, Voicetta grades the transcript against all enabled evaluations. Each evaluation returns one of three results:

- **Pass** — the expected behavior was shown.
- **Fail** — the situation occurred but the agent did not behave as expected.
- **Not applicable** — the situation never came up in the call.

Every result includes a short rationale explaining the verdict.

### AI Evaluations Page

The main menu has a new **AI Evaluations** entry. The page lets you:

- create, edit, and delete evaluations,
- enable or disable individual evaluations without deleting them,
- see aggregate pass rates and pass/fail/not-applicable counts per evaluation across your calls.

### Evaluation Results In History

Call details in **History** now include an evaluations section showing how the call scored against each criterion, with the rationale for each verdict.

## What Was Changed

### App Version

The app version badge now shows `v3.4`.

### Centrally Managed Analysis Models

The AI models used for post-call analysis (summaries, intent, sentiment) and AI evaluations are now managed centrally by Voicetta, so model upgrades reach all workspaces immediately — no app update or action on your side required.

## Important Notes

### Grading Runs On New Calls

Evaluations grade calls that end after the evaluation was created and enabled. Existing call history is not graded retroactively.

### Transcript-Based

Grading uses the call transcript. Calls without a usable transcript are marked **not applicable** rather than guessed.

## Related pages

- [App Map](/docs/app-map)
- [History and Guest Timelines](/docs/history-and-timelines)
- [Changelog](/docs/changelog)