Veai 5.7 for JetBrains IDEs adds directories to chat and automatic retries for agent tasks

Veai has released version 5.7 of its AI agent for JetBrains IDEs, focusing not on flashy demos but on everyday working scenarios. The update removes unnecessary manual steps in the chat, shortens pauses between steps and makes long tasks more resilient to random failures.

Context by whole folders

The main practical change: you can now pass not only individual files to the chat but entire directories. Previously, developers had to assemble context by hand: attach the needed files one by one, describe a module's structure in words, or hope the agent would find the neighboring parts of the project on its own. In version 5.7, you can simply drag a directory into the chat window or add it through the Add to Chat action in Project View. The agent then receives the full path and starts working with a group of related files right away.

On paper this looks like a minor UX tweak, but the effect is more noticeable in real development. When you need to analyze a subpackage, prepare a refactoring, understand a module's architecture or go through configs, the completeness of context directly affects the quality of the answer. The less manual assembly before a task begins, the lower the chance that the agent misses an important dependency, a neighboring class or a service file. For large codebases this is not cosmetic: it saves time on every cycle of interaction with the AI.

Models and resilience

Another important change: model selection now lives right in the chat. Previously, access to models from different providers was buried in settings, which slowed work down: switching models for a specific task meant extra navigation through the interface. Now the list sits in a familiar place, where it is easier to treat as part of the workflow. One menu gathers models from different providers, including variants with different reasoning modes, for example Think, if it is available for a given model.

This makes the product more flexible in daily use. A developer can switch faster between a cheaper model for routine edits, a stronger one for complex refactoring, and a reasoning mode for tasks that need depth of reasoning.

At the same time, Veai added automatic retry on agent errors. Where a background task could previously stop after the first failure and require a manual restart, the system now makes repeat attempts on its own. For long autonomous scenarios this is not a minor detail but basic reliability.

• Adding a directory to the chat via drag-and-drop or Project View
• Switching between OpenAI and Anthropic models right in the chat interface
• Support for different reasoning modes for suitable models
• Automatic retries after agent failures
• Fewer manual actions in long chains of work

An interface without the noise

Veai also went through the parts of the interface that did not break the workflow outright but irritated users every day. One example is the terminal's behavior when running commands. Previously it could open unexpectedly and grab focus, pulling the developer out of the editor while the agent was simply running a background command. In 5.7 the terminal no longer appears unless necessary and does not pull attention to itself. Users can keep writing code or reading results while the tool does its housekeeping in the background.

The team also reworked the feedback buttons in the chat. Instead of the old like/dislike pair, the interface now separates a reaction to an answer from an actual error report. Bug-report submission has moved to the top, next to the chat-saving option, so the path to a problem report is shorter. This is an important sign of product maturity: good AI tools win not only on model quality but on how quickly users can report a failure without leaving their working context.

What this means

Veai 5.7 shows a clear trend in AI development tools: the competition is no longer only about generation quality but about the friction around it. The winners will be products that cut unnecessary clicks, do not derail background tasks and let users change models quickly to fit the scenario. For JetBrains IDE users, this update looks like a set of small fixes that together make the agent noticeably more convenient in daily work.