What’s new in Rocket.Chat 7.10

In high-stakes operations, one small mistake can ripple through an entire mission.

A question phrased the wrong way, a sensitive detail slipped into an AI prompt, or an overlooked message can slow decisions or put data at risk.

That’s why Rocket.Chat v7.10 was built with teams like yours in mind.

With Rocket.Chat AI, your queries are automatically sharpened, sensitive information stays protected, and every answer is traced back to its source. The result is insights you can trust and act on immediately.

Behind the scenes, better navigation and improved visibility into apps help you cut through the noise, spot issues fast, and keep operations running without surprises.

Prompt guardrails: AI that protects as it works

AI can be a powerful ally, but without safeguards it can also introduce risk.

Even a single prompt referencing a classified project, operational plan, or regulated data can expose sensitive information.

Rocket.Chat AI already respects clearance levels and access permissions. Sensitive information is never included where it shouldn’t be, while every user still gets the insights they are authorized to see.

Prompt Guardrails is built on this foundation. Every prompt is screened before it reaches the AI.

Anything referencing classified or regulated information is flagged and blocked.
Responses that could expose sensitive data are stopped before they’re generated.
Vague or unclear questions are refined automatically so the AI gives precise, actionable answers.
All responses respect security policies to protect sensitive and classified data.

This approach addresses a key operational challenge. Even minor slips in phrasing can jeopardize classified intel, but Prompt Guardrails act as a safety net, stopping risky prompts before they become a problem.

For example, an analyst querying the status of an ongoing operational plan could inadvertently reference classified project details. Prompt Guardrails would flag or refine the prompt.

But the real power here is in the blend of protection and intelligence.

That’s why we’ve also introduced Query Augmentation to sharpen queries automatically, so AI delivers precise, actionable answers without ever revealing information it shouldn’t.

The result is teams get the insights they need with confidence.

Queries are refined, answers are clear and ready to act on immediately, and sensitive information remains fully protected.

AI stops being a potential liability and becomes a trusted partner, enabling teams to use AI confidently and securely in mission-critical scenarios.

Source attribution: AI you can trust

In high-stakes environments, you cannot act on an answer without knowing where it came from. One misinterpreted insight or unverified recommendation can halt operations, misdirect resources, or create compliance risks. This is not theoretical.

The U.S. Department of Defense has explicitly recognized the importance of traceable AI. At Rocket.Chat, we put this principle into practice.

"AI must be transparent and auditable, with clear traceability to its data sources."
~ U.S. Department of Defense, Ethical Principles for Artificial Intelligence

Source Attribution ensures that every AI-generated summary, answer, or recommendation references the specific messages or documents that informed it.

Teams can verify outputs, confirm accuracy, and act with confidence. This way, AI stops being a black box and becomes a reliable partner. This will allow organizations to navigate and make high-consequence decisions with certainty.

Here’s why it matters:

Verify with ease: check responses against original intelligence without chasing source material.
Act faster: make well-informed decisions quickly with full context at hand.
Stay accountable: maintain a complete, auditable trail of every AI-assisted operation.
Meet mandates: comply with security and regulatory standards without bending workflows to fit AI.

Bits and bytes you shouldn’t ignore

In addition to v7.10, here are a few recent AI updates already available to teams:

Streaming responses

No more waiting for the full answer to load. Now AI starts showing results within seconds, updating as it goes.

When operational data changes mid-mission, teams can catch early insights immediately. Waiting for a full summary is a thing of the past. Quick access means faster decisions, fewer bottlenecks, and less time staring at a spinning wheel.

Role-based access

Not everyone needs the same level of AI access, and that’s a good thing. Administrators can now control who can interact with AI. Sensitive tools stay in the right hands, ensuring only the authorized personnel get in.

Agent responses

For organizations using custom agentic setups, AI responses stream in real-time. Operations centers don’t have to wait for a complete summary. Instead, they can react to new intelligence the moment it arrives.

Other v7.10 updates you should know about

App log and status observability

Keeping critical apps running smoothly at scale is no small task.

Logs pile up fast, and multi-instance deployments can make it tricky to see what’s really happening.

With 7.10, Rocket.Chat introduces:

Advanced filtering by log level, method, time, or instance
Pagination and export options for deep analysis
App status visibility across all instances

This ensures operators and admins get a clear, real-time view of app health. Troubleshooting happens faster, support escalations drop, and teams can work confidently knowing their infrastructure is solid. This is significant in environments where uptime and reliability of apps can directly impact operations.

Enhanced navigation filters

High volumes of messages can overwhelm even the most disciplined teams. But finding the right conversation or channel shouldn’t slow you down. The new side navigation filters let teams:

See all conversations at a glance
Filter by Mentions, Favorites, or Discussions
Focus on citizen engagement queues

Users can prioritize critical threads, respond faster, and stay organized while cutting through the noise. In fast-paced operations, these filters help people focus on what really matters.

This feature is currently available under Feature Preview, so you can test it out and tailor it to your workflows. Each user can enable or disable the experience individually, depending on operational needs.

Enable it via: User Profile > Feature Preview > Navigation > Enhanced Navigation Experience.

Bug fixes

This release includes a number of fixes to improve overall stability, performance, and user experience. For a complete list of resolved issues in Rocket.Chat v7.6, check out the official release notes.

How to update your workspace to the latest Rocket.Chat version?

SaaS workspaces

The release happens automatically on our cloud, so no further action is required from your side. However, remember that updating instances might take a few weeks, so don't hesitate to contact our support team if you need to update your version sooner.