Customer Experience News is a weekly newsletter and video about the most important news and discussions for Customer Experience and Customer Support Leaders.
This is all the weekly news you need in around 10 minutes.
Today’s headlines are from:
Decagon Lets Teams Build Charts by Asking
Decagon, a customer service AI platform, announced Generative Analytics, an expansion of its Duet reporting tool that lets teams turn any conversation insight into a custom chart using natural language.
Support leaders can now prompt Duet to build time series, bar charts, and distributions without a data team, and new Action Trackers let teams define custom outcomes, such as completed payments or product recommendations, so those signals feed directly into shared dashboards.
Teams should plan time to define the outcomes worth tracking first, since Action Trackers only surface data that has been explicitly configured to capture it.
Guru Adds Version History to Agent Prompts
Guru, a knowledge management platform, announced Prompt and Skill Versioning for its Knowledge Agent, giving every change to an agent’s system prompt and skills a full, comparable history.
Teams that own or manage Knowledge Agents can now open a History view to see who changed what and when, compare two versions side by side, or revert with change notes, and every answer now records which prompt and skill version produced it.
Access follows the same permissions as editing the agent, so teams with audit or compliance obligations get a defensible record without changing how the agent behaves for end users.
Mintlify Adds Audio Narration to Docs Pages
Mintlify, a documentation platform, announced Narration, which lets readers listen to supported docs pages while Mintlify highlights each sentence and scrolls to keep it in view.
The feature understands documentation structure, reading titles and body copy while skipping code blocks and tables that do not translate well to audio, and it works across languages using a page’s configured locale.
Narration is available to customers on Pro plans and above and uses AI credits, charging once the first time a page version is heard, with replays free.
Crisp Adds AI Observability and a Refreshed UI
Crisp, a customer messaging platform, announced a bundle of updates headlined by Hugo Observability, which gives teams a detailed timeline of each AI-powered conversation showing which models and tools were used, how the Agent reasoned through each step, and where errors occurred.
The update also refreshes the interface across Inbox, Visitors, Contacts, and Analytics, adds reusable Operator Teams that sync membership automatically across mentions and routing, and brings Hugo access to GPT-5.6, Claude Sonnet 5, and two Gemini models.
Support leaders troubleshooting AI-handled conversations get a direct line into what the Agent actually did, rather than reconstructing it from the transcript alone.
Help Scout Adds SLAs and Company Profiles
Help Scout, a customer support platform, announced its Fall 2026 seasonal release, headlined by SLA policies that let teams set service standards with flexible conditions and route or flag conversations as deadlines approach.
The release also adds Topics, which uses AI to surface recurring themes across conversations, a Company Profile view that brings a customer’s full activity and support history into one place, native SMS support, a customer portal, and an MCP connection that gives AI tools secure access to conversation context.
Teams weighing which piece to adopt first should note that SLAs and Topics both depend on clean, consistent conversation data to produce a reliable signal.
Intercom Adds Custom Metrics and Faster Ticket Macros
Intercom, a customer service platform, announced two Reports updates. Custom metrics let teams apply their own filters to a built-in metric or divide one metric by another with different filters on each side, calculated across a workspace’s full history and available in the picker for every chart.
Separately, the ticket-creation screen now includes the full composer, so Support Specialists can apply a macro and stage follow-up actions like assigning or closing a ticket in one step instead of two.
Together the updates trim manual reporting work and cut the steps needed to close out a routine ticket.
Gorgias Lets Shoppers Confirm Before AI Handover
Gorgias, a conversational AI platform for ecommerce, announced that merchants can now choose how AI Agent hands conversations to a human teammate on Chat.
With confirmation turned on, a shopper is asked to confirm before the handover happens, rather than the handover starting right away as it did before.
Retailers running high-volume Chat queues get a way to reduce mistaken handoffs while keeping the faster, no-confirmation flow available where it works better.
Plain Speeds Up Onboarding and AI Workflow Building
Plain, an AI support platform, announced two updates. New workspaces now start with Slack-based teammate suggestions, label suggestions a team can accept or dismiss, and test threads with fake customers, so a team can check its labels, workflows, and replies before going live.
Separately, Sidekick can now build a full workflow from a plain-language description of the outcome a team wants, assembling the triggers, conditions, and actions for a person to review, adjust, and publish.
Both updates are aimed at teams standing up or reworking their support setup, cutting the time between deciding what should happen and having it running.
Tidio Adds MCP Connectors to Lyro
Tidio, a customer service platform, announced MCP connectors for its AI Agent Lyro, letting teams connect an MCP server, choose which tools Lyro can use, and let it act while a conversation is open.
Connected to Shopify, for example, Lyro can pull up a live order, tell a customer where it is, and change a shipping address before it ships, with no handoff and no ticket.
Teams choose exactly which actions are turned on, so a capability like order lookup can be enabled while something like refunds stays off until a team is ready for it.
Linear Adds New Controls for Its Coding Agent
Linear, a project management platform for teams and agents, announced new controls for its coding agent’s sessions.
Adaptive routing sends simpler tasks to a faster model automatically while reserving the default reasoning model for complex work, with the option to restart a session on the default model if a task needs more capability than expected.
Coding sessions can now access environment secrets to fetch private dependencies during setup, and workspace admins can choose which models, including open-weight models starting with GLM, are available to their teams.
GitBook Agent Now Lives in a Persistent Dock
GitBook, a documentation platform, announced that GitBook Agent now lives in a persistent dock instead of a floating panel, so it keeps working as a person moves between pages.
Anyone in an organization can now use the Agent with their own permissions, meaning it reads what they can read and edits or comments only where they can, and it says when it cannot complete a task instead of failing silently.
The Agent also draws together every connected Slack, GitHub, and Linear channel, so it can answer questions and open change requests from the same thread across every connected site.
Maven AGI Rebuilds Integrations as Full Applications
Maven AGI, an AI agent platform, announced Maven Apps, a framework built as a real application platform rather than a connector layer for its agents.
Each app is multi-tenant and isolated, and reacts to real events, such as a conversation being created, instead of running on a polling schedule, with a validate-then-run lifecycle that catches a broken credential before it reaches production.
Teams evaluating a Salesforce or Zendesk integration get a clearer picture of what they are actually getting: whether the agent can read knowledge, take action such as issuing a refund, or sit inside the console as a co-pilot, since Maven treats those as different kinds of apps rather than one bundled connector.
Pylon Ships 54 Updates Including an Integrations Hub
Pylon, a support platform, announced 54 updates in a single week, headlined by a new Integrations page that lets teams manage every connected email account and its filtering settings from one place.
The release also adds semantic search over survey responses in Ask AI, real-time source citations from AI agents during response generation, clear notifications when an MCP connector’s authentication or permissions fail instead of a silent error, and the ability to drag multiple issues at once on kanban boards.
Teams running a high volume of connected inboxes get a single place to audit and adjust email routing instead of checking each account separately.
Pluno Adds Personalized Greetings and Escalation Messages
Pluno, an AI support agent platform for B2B teams on Zendesk, announced that Messaging Deflection can now show customer-specific greetings and escalation confirmations, so a team can set clear, personalized expectations at the start of a chat and when it hands off to a human.
The release also lets teams configure a delay before a reply sends along with a typing indicator, and makes filtered views of the Reporting page shareable with the rest of a team.
Support leaders get more control over tone at the moments customers are most likely to judge it, without needing a developer to change the message.
DeepL Voice Adds Five New Languages
DeepL, a translation platform, announced Voice API support for Catalan and Galician as new source and target languages, plus translated speech output for Hebrew, Slovenian, and Thai, all in beta.
Transcription and translated speech for these languages come from external service partners rather than DeepL directly, so the source language must be set explicitly instead of relying on auto-detection.
Teams supporting customers in these languages by voice can now offer transcription and spoken replies rather than text-only translation, though the beta and external-partner status are worth noting before relying on the languages for a live queue.
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.
