Welcome back to the AIsuru release notes for March 2026! Here you'll find new features, improvements, and bug fixes to make your AI Agents more powerful and secure 🚀
NEW FEATURES AND IMPROVEMENTS!
Environmental Impact of Agents 🎉
We have now added a new integration with Ecologits in the “Conversations” page in our AIsuru platform. Within the detail of a conversation, the new option “Show message consumption” allows the user to see the consumption of each messages in the chat.
There, the user will see the data of the impact of the chat and each message, shown by energy and water used and CO2 emissions.
This feature is focused on LLMs by major providers (Mistral/OpenAI/Anthropic) and the most common open source and local models.
Additional download button
The user is now able to download their chats in a specific selected time period using the "download conversations" button in the to right corner of their "conversation" page
New fallback mechanism has been added for moderation models
Previously, if the moderation service (system that checks content for inappropriate language) had any technical issues or went down, it could have potentially blocked users from saving or performing actions on the platform.
Now, a "fallback mechanism" has been added we like to call it the "safety net". If the moderation service runs into a problem, instead of blocking everything and stopping the user from working, the system will simply let the action go through anyway, so users can continue working without interruption.
New "BANNER" notification system
A new type of notifications has now been added in a form of a banner type that can be clearly visible in the dashboard.
Hide 3D Avatar
In the layout section and more specifically under the embed panel we have now added a new feature "Hide 3D Avatar” that is only activated if only the WEBSITE_ASSISTANT layout is selected. This allows the user to hide the 3D Avatar in the bottom right of the page for better visibility.
Improvements for our models
We have now set Anthropic Claude Sonnet 4.6 and Opus 4.6 models to a 1M Tokens of context. This translates to longer documents, richer interactions, and more accurate responses when working with large volumes of content.
Also added a new OpenAI gpt5.3 instant model.
Deep thought improvements
Known facts in context now contain date and time so the system is more accurate to place the user's key point information in the timeline.
WHAT WE FIXED!
Bug fix with the filter section
In the right sidebar that opens after the user selects the filters section in the "Contents" page we fixed an issue that triggering a full reset of the filters after the user closed and reopened the window.
Error for public agents
We have resolved an issue where the agent was logging out the user and asked for a password to reenter, even if that agent was public.
🛠️ As always, we've addressed various minor bugs and implemented performance improvements across the platform to ensure a smoother, faster experience.
💡 Need help with AIsuru? Want to explore the AIsuru AI Academy? Contact us at demo@memori.ai