en

What's new in AIsuru | v10.0.0

Memori.AI

— 3 min read
Share
What's new in AIsuru | v10.0.0

Welcome back to the AIsuru release notes v10.0.0 ! Here you'll find new features, improvements, and bug fixes to make your AI Agents more powerful and secure 🚀

NEW FEATURES AND IMPROVEMENTS!

Corporate MCPs

Now tenant admins can configure MCPs for their users and agents. In the tenant settings, they will find a new section called “Corporate MCPs” where they can choose between available MCPs and configure them with credentials or settings that will be managed in a centralized way.

This way the admin users can set sensitive information while final users will be able to enable those MCPs without knowing the credentials. This allows corporate tenants to have a separation of concerns between admin users and final users among sensitive data and credentials.
Also allows admins to rotate keys and apply their security policies on every agent centrally.

Final users will find a new “Extensions” tab in the "Create" or edit section of an agent for their tenant, where they will have the list of these corporate MCPs, with no or fewer parameters.

MCP Dashboard and prompt updates

Among the connected MCPs the users can now find a new "dashboard" button for the MCPs that are able to save and process data.

When selected they will be able to manage and view their saved data. For example in the Scheduler MCP the user can view the scheduled tasks and their recurrence. While in the Persistence MCP the data that the agent has saved will be shown.

We also love to make sure that all your MCPs instructions are up to date so, as a new feature the user will be notified every time there is a new prompt version in the MCP section.

Simply click on the update bottom to get the latest prompt for our system and therefore the agent will have the best instructions on how to use them.

New asynchronous API for chat interactions

We introduced a new asynchronous API to send a message to agents: /memori/v2/TextEnteredEventAsync/{strSessionID}
This will not provide directly the generated answer, but it will allow clients to subscribe to a WebSocket from which they will receive events during the chat.

At this stage, clients receive only the completion of the answer along with any error messages. Support for progressive feedback events from tools and MCPs is currently under development.

Brand new models are now available

  • Anthropic Opus 4.7, Opus 4.8 and Claude Sonnet 5 (With Adaptive Reasoning)
  • We added Fable 5 to Anthropic supported models
  • Implemented Anthropic Adaptive Reasoning support (now for Sonnet 5 and Opus 4.7+ models)

Optimizations

  • We have increased the maximum number of requests for tools during chat interactions.
  • Now the user has more lasting timeouts for system integration webhooks.

🐞WHAT WE FIXED!

  • Fixed media (images) support for Vertex and custom models
  • Context variables creation now correctly handle special characters (such as é, ü, ñ, emoji), preventing unexpected errors
  • We've added command URL detection for custom MCPs to enhance security

🛠️ 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

Read more