# October 2025 releases > Expanded AI tooling, automation enhancements, reporting upgrades, workflow improvements, and new controls for scalable customer operations. Source: https://help.kustomer.com/en_us/october-2025-releases-BJ1_ccpnge Last updated: 2026-06-01T23:26:59.206Z Expanded AI tooling, automation enhancements, reporting upgrades, workflow improvements, and new controls for scalable customer operations. We’re continually working to improve our platform for you. This page highlights new features and improvements on the Kustomer platform, all in one place. Check here regularly to stay up-to-date with the latest changes. ### 🎉 AI Assistants are officially live! (Oct 30) We’re thrilled to announce the general availability (GA) of AI Assistants, a powerful new suite of tools designed to enhance every part of your Kustomer experience. Starting today, all organizations can access: * [AI Agent team assistant](https://help.kustomer.com/en_us/use-the-ai-agent-team-assistant-to-build-better-ai-agents-B1s6VhVebl) * [AI workflow assistant](https://help.kustomer.com/ai-workflow-assistant-S1S1EG_3ee) * [AI search assistant](https://help.kustomer.com/ai-search-assistant-HyNTVMkhlg) * [AI knowledge base assistant](https://help.kustomer.com/ai-knowledge-base-assistant-SkfK37k2eg) Each Assistant brings unique, intelligent capabilities that help teams work smarter and deliver exceptional customer experiences. From more capable AI Agents and streamlined automations to richer insights and smarter content, AI Assistants elevate your operations across the entire platform. **Bonus:** This release also introduces **speech-to-text support** across all Assistants, **Copilot**, and the **Timeline reply editor** — making it easier than ever to work hands-free. **Who has access?** All Kustomer clients now have access to AI Assistants. Permissions for each Assistant follow the same rules as the feature they support. For example, if a user doesn’t have permission to edit a Search, they won’t be able to do so through the AI Search Assistant. ### 🚀 Introducing the Kustomer MCP Server (Oct 30) We’re excited to announce the launch of the Kustomer MCP (Model Context Protocol) Server — a new way to securely connect Kustomer to large language models (LLMs), such as Claude and ChatGPT. The MCP Server provides structured, monitored access to your Kustomer data, allowing AI systems to retrieve and summarize information directly from your instance using predefined tools and secure API endpoints optimized for AI use. This feature is available to all organizations on all plans. Learn more in our help articles: * [What is the Kustomer MCP Server?](https://help.kustomer.com/en_us/kustomer-mcp-server-r1CR1NpTgx) * [Set up the Kustomer MCP Server](https://help.kustomer.com/en_us/kustomer-mcp-server-HJsPzR56ee) * [Example use cases for the Kustomer MCP Server](https://help.kustomer.com/en_us/kustomer-mcp-server-ryXMRehTgl) ###  🧠 Evals Update: AI Agent Assertions (Oct 29) Admins can now define expected AI Agents to be used for a given test query—similar to how tool use is evaluated. During evaluation, Evals will check whether the correct agents were invoked as intended when addressing the query. This enhancement helps ensure multi-agent workflows are functioning as designed and that the right agents are engaged for the right tasks. This feature is available for organizations with access to AI Agents for Customers. ### 📝 Summaries can now be enabled/disabled by category (Oct 27) When you’re configuring [AI Summaries](https://help.kustomer.com/en_us/set-up-summaries-BkESHdVmee) for your agents, you can now individually enable or disable summaries by category. This includes: * Conversation summaries * Customer summaries * Conversation handoff summaries * Conversation closing summaries This feature is available for organizations with access to AI Agents for Reps. ### 🔒 PCI redaction is now generally available (Oct 24) With this update, you can now enable [PCI redaction](https://help.kustomer.com/en_us/pci-redaction-SkQQAKARgx) through the **Sensitive Data Protection** option in your Security settings. When enabled, any credit card information that’s accidentally shared in supported channels and outgoing conversations will automatically be redacted and replaced with X’s. **Note:** This feature does not make Kustomer a PCI-DSS compliant payment processor or approved place to store credit card numbers. Instead, it helps your organization stay PCI Compliant by being Out of Scope by preventing Kustomer from processing or storing Credit Card Numbers. This feature is available to all customers, regardless of plan. It’s turned off by default and must be enabled by an **admin** or **organization owner**. After enabling, it can take up to 4 hours for PCI redaction to take effect across your organization. ### 🪧 Identify Sandbox environments with a banner (Oct 23) When you’re logged into a sandbox, you’ll now see a banner at the top of the screen. This banner makes it easy to tell when you’re working in a **sandbox environment** (for testing) rather than a production org. The banner displays the **name of the related production org** and includes helpful information about **sandbox best practices**, so you always know exactly where you’re working. ![](https://cdn.kustomerhostedcontent.com/media/570fad9d9001bc1000163b28/a880e8d9324efe08eb646d08ae2731d1.png) ### 🧱 View Sandboxes in Settings Org Owners and Admins can now **view all sandboxes related to your production org** directly from **Settings**. This update follows foundational work to establish parent-child relationships between production orgs and their associated sandboxes. ![](https://cdn.kustomerhostedcontent.com/media/570fad9d9001bc1000163b28/727241ae005957d0bbd7a0b972bbac2d.png) This makes it easier to understand your org’s sandbox setup at a glance, all in one place. ### 🛠️ Add and remove skills on user profile (Oct 14) Building on the October 10 release, admins (or users with permission to edit other users) can now **add and remove skills** directly from a user’s profile page. This provides a more intuitive way to manage which skills a user has for routing purposes—right where you’d expect to make those changes. ![](https://cdn.kustomerhostedcontent.com/media/570fad9d9001bc1000163b28/08a131daadd365879e3abee0051bf7e8.png) This is available for organizations on the Ultimate plan using Skills-based routing. ### 📞 Kustomer Voice routing improvements (Oct 13) We’ve improved Kustomer Voice with a new routing architecture. This new architecture: * Makes it much less likely that a call will fail to route to an agent * Gives us better visibility into the few cases where something does go wrong, so that we can more easily troubleshoot and address these issues proactively * More reliably identifies agents who are experiencing issues that prevent them from successfully handling voice calls, so that they’re aware and can troubleshoot * ‘Fails faster’ - when a routing attempt does fail, it fails more quickly, so that the call can go to the next available agent faster and get answered more quickly as well **How to ensure the upgrade is fully applied** If you were using Kustomer Voice prior to October 13 and left at least 1 tab open with Kustomer, refreshing the browser tab after the upgrade will apply the upgrade. (If you didn’t have Kustomer open in a tab when the upgrade went live, you don’t need to do anything.) ### 🧩 Skills on user profile (Oct 10) We’re continuing to enhance the experience for clients using Skills-Based Routing. Admins can now **view which skills are associated with a user** directly from their profile in **Settings**. This provides quick visibility into an agent’s routing qualifications without needing to navigate elsewhere. ![](https://cdn.kustomerhostedcontent.com/media/570fad9d9001bc1000163b28/decadb9d00dd002dd2f401806a866bdf.png) This is available for organizations on the Ultimate plan using Skills-based routing. ### 🚀 Evals for AI Agent Teams are now live! (Oct 8) We’re thrilled to announce the launch of [Evaluations (Evals) in AI Agent Studio](https://help.kustomer.com/use-evaluations-to-measure-ai-performance-BJy49yooxe) — a powerful new way to measure, validate, and improve your AI’s performance before it meets real customers. AI is transforming how businesses interact with their customers, and ensuring that these systems behave reliably, accurately, and consistently is more important than ever. That’s where Evals come in. Evals let you test your AI agents in realistic, simulated scenarios, from checking delivery statuses to canceling orders, and see exactly how your agents perform. Each evaluation run automatically grades the agent’s behavior and provides detailed results, helping you spot strengths, weaknesses, and opportunities for improvement. With configurable benchmarks, automated grading, and scalable validation (up to 100 runs per scenario!), you can confidently iterate on your AI setup, catch issues early, and continuously improve agent quality. ### 💡 Highlights: * **Configurable benchmarks** – Define success with custom test cases that mirror your business workflows. * **Automated grading** – Get instant feedback with pass/fail outcomes and detailed explanations. * **Scalable validation** – Run hundreds of tests in minutes to spot inconsistencies or hallucinations. * **Actionable insights** – See exactly how each eval plays out — including errors, tool usage, and timing — so you can focus your tuning efforts where it matters most. * **Continuous improvement loop** – Regular evals help you track progress, increase reliability, and build customer trust over time. With Evals, you’re not just testing your AI — you’re building confidence that it will perform the right way, every time. See a tutorial [here](https://help.kustomer.com/run-an-evaluation-on-your-ai-agent-team-rJV8JzQnxg).  This feature is available to all organizations with access to AI Agent Studio. ### 📞 New ‘who ended the call’ notifications in Kustomer Voice (Oct 3) We’ve added notifications to give your agents more context when handling calls via Kustomer Voice. ![](https://cdn.kustomerhostedcontent.com/media/570fad9d9001bc1000163b28/6560c88ee0724e6c2e73c24c851a23c6.png) * When a call ends, agents can see who ended the call - the agent or the customer. * In cases where the call ends unexpectedly, the agent can now determine whether the customer ended the call intentionally, it was a mistake on the agent’s part, or something else. These notifications show up for all users with access to Kustomer Voice. ### 🔍 AI Agent Fallbacks in Observability Traces (Oct 3) We’ve expanded on recent improvements to [AI Agent fallbacks](https://help.kustomer.com/handle-failed-ai-agent-conversations-SkdnrD8Dxe). A few weeks ago, we introduced the ability for admins to configure fallback options in case an AI Agent encounters an error. Now, these fallback events are also reported directly in [Observability traces](https://help.kustomer.com/understand-observability-B1CMeXm1ye). This enhancement makes it easier for admins to: * Identify when a fallback occurred * Understand the cause of the error * See what action was taken as a result This update is available to all admins with access to **AI Agent Studio** and **Observability**. ### 📱 New iOS, Android, and React Native releases (Oct 2) View the [iOS](https://developer.kustomer.com/chat-sdk/v2-iOS/changelog/ios), [Android](https://developer.kustomer.com/chat-sdk/changelog/android), and [React Native](https://developer.kustomer.com/chat-sdk/changelog/react-native) changelogs for more information on what's included in these releases.