Customer Experience News is a weekly newsletter about the most important news and discussions for Customer Experience and Customer Support Leaders.
Subscribe to the LinkedIn Newsletter for the condensed version. Visit CX-News.com to research the operational impact and implementation considerations for each update.
This is all the weekly news you need in around 10 minutes.
Fin Adds Skill-Based Routing and Wrap-Up Protection
News Summary: Fin, Intercom’s AI customer service platform, released several operational updates including skill-based routing that directs conversations to the best-qualified available teammate, protected wrap-up time after chat and email conversations to let agents finish notes before the next assignment arrives, and per-teammate office hours in reporting. The company also updated the Customer Portal with full-content search across replies and messages, a conversations view alongside tickets, and segment-based visibility controls.
Operational Impact: Skill-based routing fills a gap in balanced assignment: conversations were distributed by workload, but not by expertise. Admins can define skills such as billing, product area, or region and route conversations to the best-matched available teammate, with standard load balancing as a fallback when no match is available. Protected wrap-up time addresses a common pressure point where agents had to choose between completing post-conversation work and accepting the next assignment. The held slot opens for a configurable window per channel before new work lands. Per-teammate office hours means first-response metrics are now measured against each agent’s actual shift, which matters most for distributed teams where a Support Specialist in one time zone was being measured against a workspace default set for another.
Implementation Considerations: Skill-based routing requires upfront admin configuration: defining the skill taxonomy, enabling the feature per team inbox, and mapping teammates to their skills. A poorly defined taxonomy creates routing gaps as easily as standard load balancing does. Wrap-up protection is off by default and must be enabled per channel; teams should review existing workload management settings before turning it on to avoid conflicts. Per-teammate office hours requires individually assigning schedules from the new Teammate Schedules tab; without this step, reporting falls back to team or workspace defaults, which may appear correct but will not reflect actual agent availability.
Zendesk Releases Predictive Routing, Free WFM, and Contact Center GA
News Summary: Zendesk, a customer service platform, released its July 2026 product updates including AI-based predictive routing in omnichannel queues, which assigns messaging tickets to the agents predicted to resolve them fastest, and the general availability of Zendesk Contact Center integrated directly into Agent Workspace. Workforce Management is now included at no extra cost for Suite Professional plans, and five new action builder connectors — HubSpot, Chargebee, Stripe, Calendly, and GitHub — are available for automation flows.
Operational Impact: Predictive routing adds an AI layer on top of standard queue assignment: rather than distributing by availability or round-robin, tickets go to the agent whose resolution history best matches the incoming request. A new Agent Engagement Time metric tracks agent activity to help teams measure whether the routing model is improving outcomes. The Contact Center GA consolidates phone support into the same workspace used for digital channels, removing the interface switch between voice and ticket work. WFM access on Suite Professional without an add-on fee lowers the adoption barrier for teams that have deferred workforce management due to cost; the included features cover agent activity visibility and daily average handle time. The QA Coaching dashboard connects coaching sessions, feedback reviews, and IQS impact in one view.
Implementation Considerations: Predictive routing is only available in omnichannel routing configurations; teams using rule-based or manual routing will not see the option. The WFM access on Suite Professional covers 24 hours of activity data; teams that need historical lookback for staffing models will need the full WFM add-on. Connecting external knowledge sources such as Google Drive or Amazon S3 will surface existing content quality problems; a knowledge audit before connecting is worthwhile since AI agents and generative search will draw from whatever is connected. The five new action builder connectors require separate authentication and should be tested against each org’s specific instance before use in live automation flows, particularly for billing or scheduling actions that carry customer-facing risk if they fire incorrectly.
Notion 3.6 Adds External Agents and Microsoft File Support
News Summary: Notion, a productivity and workspace platform, released version 3.6 adding External Agents including Claude and Cursor that can be assigned tasks from a shared board and run alongside human teammates, and support for PPTX, XLSX, DOCX, and PDF files for Notion Agents. The release also added calendar tools for the AI agent, a dedicated Notion Agents iOS app, and the ability to share Notion Workers across a team, along with five new MCP connections including Mercury, Mixpanel, and Miro.
Operational Impact: External Agents make AI agent activity visible and assignable from the same boards used for human work, rather than running in isolated sessions no one else can see. For CX operations teams using Notion as a project hub, this means support process documentation, triage rules, and roadmap feedback from customers can be handed off to automated agents using the same task structure used for human assignments. Microsoft file support extends Notion’s usefulness in enterprise environments where compliance reports, runbooks, and briefings live in Word or Excel; agents can now read those formats for context and generate them as outputs. Shared Notion Workers allow a single integration or automation built by one team member to be made available to the rest of the workspace, which matters for CX teams that want engineers and product managers to receive structured bug reports or feature signals without manual handoffs.
Implementation Considerations: External Agent features for Claude and Cursor and some Notion Workers capabilities are in early access or alpha; teams should verify current availability before building workflows around them. Notion Workers will move from free beta to Notion credits billing starting August 11, 2026; organizations building on Workers during the free period should model credit costs before that date. The Custom Agent audit log, which gives compliance and security teams visibility into agent activity, is available on Enterprise plans only. Notion AI credits consumption will increase with Agents iOS and scheduled agent features; admins should set per-agent credit limits before enabling broad team access.
Pylon Relaunches as an Agentic Support Platform
News Summary: Pylon, a B2B customer support platform, announced it is repositioning as an agentic support platform, replacing its deflection-first model with AI Background Agents that investigate issues, gather context, and draft responses using reusable natural language Skill instructions. The company launched with 20 early access customers and reports that one customer reduced L1-to-L2 escalations by 70% during the program.
Operational Impact: The Background Agent model changes the role of AI in the support queue from answering FAQs to performing work. Rather than generating a deflection response when a ticket arrives, the agent investigates the issue, checks relevant systems and conversation history, and prepares a draft for a Support Specialist to review or send. Skill instructions are written once and applied consistently across all agents, which addresses a gap in large teams where individual agents handle similar issues differently. The 70% escalation reduction cited is from one early access customer and should be treated as directional; the most likely candidates for meaningful reduction are teams with high escalation rates caused by context gaps rather than genuine technical complexity.
Implementation Considerations: Background Agents require that customer data, product systems, and ticket history be integrated and accessible to the agent. Teams with siloed systems or inconsistent data quality will see limited benefit from investigation steps that pull from poor sources. Skill instructions are written in natural language, which lowers the setup barrier but introduces variability; vague or overlapping Skills will produce inconsistent agent behavior. The product is in early access with 20 customers and is not yet in general availability; teams interested in this approach should contact Pylon directly and plan for an evaluation period before committing to a deployment timeline.
Sierra Launches Horizon for Long-Horizon Agent Goals
News Summary: Sierra, a conversational AI platform for enterprise customer service, announced Horizon, a platform that enables agents to pursue goals spanning days, weeks, or months, such as scheduling a specialist referral across a patient, a specialist, and a referring physician, or completing a loan origination. Horizon adds a context engine and long-horizon planning to Sierra’s Agent OS and shifts to outcome-based pricing where customers pay per business outcome delivered rather than per token consumed.
Operational Impact: Horizon extends AI agent scope beyond a single conversation window. The context engine retains what was learned in each interaction and uses it to inform the next outreach, whether that is a text, a call, or an email, without requiring the customer to re-explain their situation. For CX operations teams, the most direct use case is customer retention: the platform can run a multi-touch save play over several days when a customer signals intent to leave, rather than triggering a single automated response. The outcome-based pricing model transfers the risk of token cost management to Sierra, which removes one of the more unpredictable expense lines in AI deployment.
Implementation Considerations: Horizon is designed for enterprises managing complex, multi-step customer journeys that play out across multiple interactions over time. It will not deliver meaningful value for teams whose support interactions resolve in a single session. The context engine’s value increases with the volume of past interactions, so early performance will be lower than long-term performance and should be factored into any evaluation timeline. Teams considering Horizon need a clear definition of which business outcomes they are optimizing for before deployment, since the outcome-based pricing model requires agreeing on what counts as success. Horizon is a new product and the outcome-pricing model has not yet been proven at scale across varied industries.
Linear Adds Status, Priority, and Labels to Initiatives
News Summary: Linear, a project management platform, released Initiative Properties, adding proposed and canceled statuses, priority levels, and labels to its Initiatives feature, which organizes high-level company goals and the projects that contribute to them. The release also introduced Enterprise-managed authorization for Linear’s MCP server in Claude, allowing Okta administrators to centrally manage employee access without requiring individual authentication per user.
Operational Impact: For CX operations teams that track bug reports and feature requests through Linear, Initiative Properties adds the organizational layer needed to distinguish active roadmap bets from speculative or discontinued ones. Labels allow grouping by product line, support category, or customer segment, making it easier to surface which initiatives are relevant to a support leader’s area without scanning the full roadmap. Priority adds a standard signal that support leaders can use to communicate urgency to product and engineering teams when surfacing customer-driven initiatives. The Enterprise MCP server authorization removes the friction of individual credential setup when deploying Linear as an MCP tool through Claude.
Implementation Considerations: Initiative Properties add value primarily for teams already using Initiatives in Linear to structure roadmap planning. Teams using Linear for issue tracking only will not benefit from this release. Labels should be standardized across the workspace before broad adoption; inconsistent labeling in initiatives creates the same navigation problems it does in issues. The Enterprise MCP server authorization requires Okta as the identity provider and Claude as the MCP client; organizations on other identity providers or using other AI tools cannot use this feature.
Guru Launches Operator Mode for Knowledge Agent Configuration
News Summary: Guru, a knowledge management platform, launched Operator Mode, a chat-based interface for configuring and improving Knowledge Agents without navigating settings screens. Agent owners can describe what they want in plain language, ask the agent why it gave a specific answer, fix the underlying skill in the same session, and verify the correction by regenerating the response immediately, with all proposed changes requiring explicit approval before going live.
Operational Impact: Operator Mode compresses the feedback loop from bad answer to corrected answer. Without it, a Support Specialist who notices an agent gave a wrong answer needs to identify which skill misfired, find it in settings, edit it, and wait for changes to propagate. With Operator Mode, the agent owner can start from the bad answer itself, ask why it happened, and fix the root cause in a single session. Changes are live immediately after the owner approves, and the corrected response is available to the next person who asks automatically. Access is restricted to users with the Manage agent settings permission, which limits uncontrolled edits.
Implementation Considerations: Operator Mode is available to all Guru customers now, but its usefulness depends on the existing skill structure. Agents with vague or overlapping skills will produce explanations that are hard to act on, and a fix to one skill may expose a problem with another. Teams that have not yet organized their Knowledge Agent skills will benefit from doing so before relying on Operator Mode as a debugging tool. Operator Mode sessions are private to the user running them; changes go live upon approval but there is no built-in peer review step. Teams with quality control requirements around knowledge base changes should add a manual review process outside of Operator Mode.
Maven AGI Launches Charters to Drive Revenue from Support Conversations
News Summary: Maven AGI, an AI customer service platform, announced Charters, a feature that gives agents structured instructions to trigger revenue actions including upsell, cross-sell, nurture, and retention offers within the same governed support conversation. Charters pair with Intelligent Fields, which read live conversation context to assess customer intent, eligibility, sentiment, and journey stage, and the platform carries 15 certifications and assessments including ISO 42001 for AI governance and PCI-DSS 4.0 Level 1.
Operational Impact: Charters allow a support interaction to carry a commercial outcome without routing the customer to a separate sales flow or requiring a handoff. A retention charter can detect signals of intent to leave and initiate a save play before the customer disengages; an upsell charter can surface a relevant offer when Intelligent Fields read product fit. For support leaders being asked to demonstrate revenue contribution from the CX function, Charters create a mechanism for that without requiring separate tooling or a parallel sales system. The compliance framework backing the feature addresses the governance concern of running commercial actions through an AI agent in regulated industries.
Implementation Considerations: Charters should only be deployed after resolution quality is established and trusted. Introducing a revenue play into a support conversation where the agent is already unreliable compounds the problem: a frustrated customer who receives a misplaced upsell offer is more likely to leave than one who received a clean resolution. Intelligent Fields depend on accurate data from integrated systems; a CRM or subscription database with incomplete or stale customer records will produce misfired Charters. The results cited in the announcement are for resolution quality and are not specific to Charter-driven revenue outcomes; Charters is a new feature and does not yet have published performance data.
Pluno Launches Browser Extension for API-Direct Task Execution
News Summary: Pluno, an AI support agent for B2B teams, launched a browser extension that connects to each web application’s API directly rather than navigating screen elements, enabling it to complete cross-tool tasks across Notion, Slack, HubSpot, and other platforms without breaking when a UI element moves. The extension comes pre-trained on more than 50,000 tools, requires explicit user approval before any action runs, and is available in the Chrome Web Store; the same release added Linear support to Pluno’s Escalation Copilot with full feature parity to Jira.
Operational Impact: API-direct execution makes browser automation more reliable than screen-clicking agents in environments where UIs change frequently due to product updates or A/B tests. For support teams, this is most relevant for cross-tool tasks that currently require manual steps: updating records across a CRM and a helpdesk after a ticket closes, pulling usage data from multiple platforms for a weekly report, or reassigning contacts in bulk after a team restructure. The addition of Linear to the Escalation Copilot is a practical improvement for teams where engineering issues are tracked in Linear rather than Jira; Support Specialists can now create, link, and manage Linear issues from Zendesk and Intercom sidebars with the same workflow used for Jira.
Implementation Considerations: The browser extension requires Chromium-based browsers such as Chrome or Edge; teams on Safari or Firefox cannot use it. The pre-trained model covers more than 50,000 tools, but custom or internal tools require users to demonstrate workflows before Pluno can learn them, adding setup time. User approval is required before every action, which is a necessary safeguard but creates a workflow interruption for high-volume or repetitive tasks. Organizations evaluating Pluno should test their specific tool stack before committing, since API coverage and reliability varies across platforms.
Help Scout Adds Native WhatsApp to the Shared Inbox
News Summary: Help Scout, a customer support platform, launched native WhatsApp support, automatically routing inbound customer WhatsApp messages into the shared inbox as standard conversations that can be assigned, tagged, and managed alongside email and other channels. The integration supports text, photos, videos, voice messages, and documents inline in the conversation thread, includes template support for the 24-hour WhatsApp messaging window, and is available on all paid Help Scout plans.
Operational Impact: WhatsApp is the channel where customer inquiries most commonly land outside the support tool, either on a personal device or in a standalone WhatsApp Business account disconnected from the helpdesk. Bringing it into the shared inbox means assignment, tagging, workflows, and reporting that already exist for email apply to WhatsApp conversations without separate configuration. Template support built into the integration allows teams to send pre-approved follow-up messages after the 24-hour window closes without switching applications. All media types appear inline in the conversation thread, which matters for support flows where customers send photos or videos as evidence of product issues.
Implementation Considerations: Connecting WhatsApp to Help Scout requires an existing WhatsApp Business Account. Teams without one will need to complete Meta’s business verification process before setup, which can take several days. Templates must be created and approved in the WhatsApp Business Account before they can be used in Help Scout; teams should not wait until they need to send a template to create one. Advanced routing and automation for WhatsApp conversations are available on Plus and Pro plans; Standard plan customers will have WhatsApp in the inbox but limited automation options.
Plain Launches Sidekick Home for AI-Powered Support Investigations
News Summary: Plain, a B2B customer support platform, released Sidekick Home, a dedicated workspace in the sidebar where support teams can start AI-powered investigations, monitor all running sessions in one place, and manage integrations from a single page. Sessions can be triggered from individual threads, keyboard shortcuts, Slack, or automated workflows including scheduled recurring tasks like daily digests of overdue threads or weekly reports on support themes, with all sessions showing live status regardless of how they were started.
Operational Impact: Sidekick Home centralizes AI investigation activity that was previously scattered across individual threads. A Support Specialist or operations lead can now see every active and completed Sidekick session in one view, including sessions triggered by automated workflows they did not start themselves. Scheduled workflows extend AI capabilities to recurring operational tasks without requiring manual initiation: a workflow set to run each morning can identify SLA breaches and threads that have waited longest and post a digest to Slack before the team’s first ticket review. Integration management from the Sidekick Home page means that CRM, issue tracker, and observability connections configured once are available to all sessions, and investigations can cross-reference across all connected tools.
Implementation Considerations: Sidekick Home and scheduled workflow capabilities require Plain AI to be enabled on the workspace. Teams on plans without Plain AI will not see the feature. Scheduled workflows consume Sidekick credits; teams setting up daily automations across a large workspace should monitor credit usage after enabling to avoid depleting a credit allocation unexpectedly. Sidekick investigation quality depends on the breadth of integrations connected; teams with fewer connected tools will see narrower results from open-ended queries. Setting up workflow triggers requires configuring Plain’s workflow builder, which is a separate step from enabling Sidekick and may require admin access.
Kustomer Adds Human-Only SLA Tracking and a Major Reporting Overhaul
News Summary: Kustomer, a CRM-based customer service platform, released a large batch of updates including Skill-Based Routing in Workflows, which lets workflows add and remove skills on conversations and use skill state as a routing condition, and a new SLA setting that measures first-response time from the first human agent reply rather than from any automated reply. The release also included an extensive set of reporting improvements including a live custom chart builder and drilldowns from rate metrics into contributing records, alongside a full overhaul of mobile UX and Command Center.
Operational Impact: SLA first-response tracking from human replies fixes a measurement problem that inflates team performance data: automated first replies make response times look faster than a human actually engaged with the customer. With the new setting, the SLA clock starts when a Support Specialist responds, which is a more accurate reflection of actual customer experience. Skill-Based Routing in Workflows extends routing logic into automation; skills can be added or removed by workflow conditions before assignment runs, which enables precise matching without requiring agents to manually update skill assignments. The custom chart builder with live preview resolves a specific reporting gap: operations teams could see aggregate metrics but had to leave the report to identify which conversations drove a particular rate, and can now drill into contributing records directly.
Implementation Considerations: SLA first-response from human replies is configured at the SLA level; teams will need to update existing SLAs individually to apply the new measurement approach. Changing this setting will alter historical comparisons, so teams should document current baselines before switching. Skill-Based Routing in Workflows requires that skills-based routing already be configured in the account; teams not currently using skills-based routing will need to set it up before the workflow conditions become available. Many of the reporting improvements are in beta access, which means availability varies by account; teams should verify which specific features are accessible before building reporting processes around them.
Now this is:
Strategic Support for CX Leaders.
You’ve got ambitious Support targets and new metrics but you’re not sure what to prioritize first.
The list is long, the queue is getting longer, and you don’t have time to step back and think about CX strategically.
What if you could pressure-test your thinking with someone who’s spent 20 years building Customer Support operations?
No pitch, just a conversation.
