BTC AI Assistant 26.1.2 is available. This update makes the Assistant extensible through custom skills, adds a built-in workflow for requirements-based testing, speeds up generation, and introduces FAME — feedback-driven memory, so the Assistant can build on experience instead of starting from zero every time.
Here's what's new.
Configurable skills and a built-in RBT workflow
BTC AI Assistant can now load custom skills from a configurable location, so you can extend it with the workflows your team actually uses. The release also ships a built-in Requirements-Based Testing (RBT) workflow, which guides the Assistant through the steps of a structured testing task instead of leaving it to a single open prompt.
Improved performance
The new skill-based approach also improves overall performance. Workflows are handled in a more focused and efficient way, resulting in faster responses and a smoother experience — especially for larger or more complex engineering tasks.
Better test-case generation
Test-case generation is faster, more controllable and easier to work with:
- Project scopes — target generation at the part of the project you are working on.
- Draft and executable states — review generated test cases as drafts before promoting them to executable.
- Context is kept when you resume — if generation pauses for a clarification, your answers are retained rather than lost.
- Generated test cases get unique default names, so nothing collides in a larger project.
Learning from experience with FAME
FAME — the Feedback Adaptive Memory Engine — introduces feedback-driven memory for BTC AI Assistant. Initially focused on test-case generation, it can retain selected outcomes and user feedback as reusable experience, and retrieve relevant lessons for similar future tasks. This is designed to help reduce repeated mistakes and improve consistency over time, while keeping the memory external to the selected LLM and scoped to the configured environment.
Clearer progress, more reliable runs
A new active-jobs view shows background operations directly in the chat, so you can see what the Assistant is working on rather than waiting on a silent window. Progress messages during long-running workflows are clearer, and formalization handling has been improved for more consistent results.
This matters most for the long-running tasks. A requirement proof can take a while, so the Assistant starts it in the background, reports that it has started, and keeps the chat usable — with the job and its status visible in the sidebar throughout.
How to get BTC AI Assistent 26.1.2
BTC AI Assistant is delivered as a plugin, installed and updated with the BTC Plugin Manager. Download the Plugin Manager application once from the MyBTC customer portal, then open it to install version 26.1.2 — and to keep your plugins up to date from then on.
New to BTC? Request a free evaluation license.
New here? What BTC AI Assistant does
BTC AI Assistant brings AI-assisted workflows to BTC TestStack and BTC EmbeddedPlatform. Through its chat interface, you can ask questions about the product, work with supported engineering artifacts, and carry out tasks using the tools available in your current environment. Depending on your project and configuration, it can also access relevant work products such as test results and code-coverage metrics.
Supported capabilities can also be exposed through an MCP server, enabling agents and other MCP-compatible AI environments to use BTC TestStack tools and workflows.
You are not tied to a single model. BTC AI Assistant supports different LLMs and deployment options, allowing you to choose where model processing takes place and align the setup with your security, privacy, and data-governance requirements.
