.png?sv=2026-02-06&spr=https&st=2026-07-22T00%3A37%3A22Z&se=2026-07-22T01%3A22%3A22Z&sr=c&sp=r&sig=tJUfRz8l97sDWYrQEA4PjE0rv884IJOhh2NIbTzDR%2BA%3D)
July 15, 2026
Signals: Evaluate AI Conversation Quality at Scale
Signals is a new way to evaluate the quality of your AI Sessions at scale – no surveys, manual review, or third-party tools required.
Define what matters – Write evaluation criteria in plain English around the things most important to your operation: sentiment, policy adherence, Customer feedback on a new product change, or anything else you want to track.
Run at scale – Evaluate hundreds of Sessions at once and see pass/fail breakdowns, giving you continuous visibility into how your AI is performing at a scale no team could match manually.
Drill into any result – Click into any evaluation result for the full transcript and AI reasoning behind the pass/fail decision.
Monitor and uncover insights – Track AI Sessions and surface Customer insights, directly within Gladly.
Signals is available for free for a limited time, with paid pricing expected later this fall. Using Signals during the free period does not obligate you to purchase it once pricing is introduced.
To get started, contact your Strategic Account Manager or Gladly Support to have Signals enabled for your organization.
Improved Post-Call Summaries
Voice summaries are now delivering better accuracy and richer context. Filler phrases like "the transcript" or "the recording" have been replaced with specific details Agents need to pick up a Conversation or follow up, without re-reading the transcript.
Summaries also follow a new structured format instead of a single dense paragraph. Rather than scanning a block of text, Agents get four clear sections:
Issue – What the Customer contacted you about.
Key Points – Specifics from the call (order numbers, account details, dates).
Outcome – Whether the issue was resolved, and how.
Next Steps – What (if anything) still needs to be done.
If your organization already uses Voice summaries, this improvement applies automatically with no action required.
"Conversations" Renamed to "Sessions" in Gladly AI
"Conversations" has been renamed to "Sessions" throughout the Gladly AI interface – most visibly on the "Gladly AI Conversations" page, which is now "AI Sessions." The update reflects how these terms are already defined in the Gladly platform:
Conversation – The time-bounded grouping of all interactions a Customer has with your brand. It can include any combination of AI Sessions and team-handled Contacts.
Session – A single Gladly AI interaction – one Chat, one phone call, or one Email thread handled by AI. It stays open while the Customer and Gladly AI are actively exchanging messages and closes after 2 hours of inactivity.
Because Gladly AI operates at the Session level – handling one interaction at a time – the UI now reflects that distinction. This is a terminology change only; no features, data, reporting, or billing are affected. Learn more about Conversations vs. Sessions – and which reports to use for each.
Redesigned Knowledge Sources Modal for Guides
We’ve refreshed the Knowledge Sources action configuration for Guides. Answers, Websites, and Documents categories are organized into tabs, and documents can be selected from a checkbox list (with select-all and deselect-all) instead of individual on/off toggles. This makes it easier to see at a glance what information a Gladly Agent has access to and will search when generating a reply.
Inline Customer Images in AI Sessions Review
When reviewing Gladly AI Sessions, images that Customers sent appear inline on the AI Sessions page with click-to-enlarge, instead of showing only the file name. Reviewers are able see the image the AI was responding to and give more complete feedback.
"Starts With" and "Ends With" Condition Operators for Workflows
Workflow authors can match text with two new condition operators – "Starts with" and "Ends with" – for cleaner prefix and suffix rules. For example, route on order numbers that start with a given prefix, or match email addresses that end with a specific domain. Previously only "Contains" was available, which matched the text anywhere in the value.
Gladly AI for Helpdesks: Improved Tagging on AI Responses
Two improvements for organizations using Gladly AI within an external helpdesk:
Explicit AI outcome tags – Tickets created in your external helpdesk are tagged with a clear AI outcome:
gladly-ai-handoffwhen Gladly AI hands off to an Agent, andgladly-ai-resolvedwhen Gladly AI resolves the Conversation.Abandoned Chats tagged separately – Chats the Customer abandoned or that timed out are now tagged
gladly-ai-abandoned, so your helpdesk reporting can distinguish an abandoned Chat from one Gladly AI actually resolved.
.png?sv=2026-02-06&spr=https&st=2026-07-22T00%3A37%3A22Z&se=2026-07-22T01%3A22%3A22Z&sr=c&sp=r&sig=tJUfRz8l97sDWYrQEA4PjE0rv884IJOhh2NIbTzDR%2BA%3D)
Agent Availability Report
A new Agent Availability report is available – a lightweight version of the Agent Summary report containing Agent identity and assignment columns plus Available Time broken out by Any, Voice, Messaging, Mail, and Task. Designed for fast retrieval via the Generate Report API, especially for workforce-management use cases with large Agent volumes.
To access, navigate to Reports → Agents → Agent Availability.
Gladly AI Feedback Report
The new Gladly AI Feedback report gives Admins and Team Managers a row-by-row export of every thumbs-up or thumbs-down review – and any comment left with it – submitted on Gladly AI Sessions. Unlike rolled-up summary reports, this report returns one row per review, so you can read individual reviewer comments and open the exact Session each review is tied to.
Navigate to Reports → Gladly AI → Gladly AI Feedback Report to either view the embedded dashboard, or download the CSV.
Search Results Navigate Directly to the Matching Item
Clicking a search result in Gladly Team navigates directly to the specific item that matched the search. Previously, clicking a universal search result opened the Conversation at the bottom, requiring you to scroll to find the match.
Table Support in Rich-Text Composers
Agents can paste tables into Note, Task, and Email Composers. Background color, text color, font weight, and text alignment are preserved on paste. A table editing toolbar appears when working inside a table, and wide tables scroll horizontally.
Additional Read-Only Endpoints Resource
Gladly's public API now exposes a read-only Endpoints resource. Developers can call GET /api/v1/endpoints – and GET /api/v1/endpoints/{id} – with their API token to retrieve the inbound addresses configured for their organization – each with its ID, address, and channel.
This makes it easier to programmatically discover which Channels and addresses an organization is set up to receive Customer Contacts on.
Unified Response Generation: AI Gets Full Context Before Responding
Today, Gladly AI handles navigation decisions (moving to another Guide section, going back, or handing off) and response generation as two separate steps. Unified Response Generation combines these into a single step, so that AI has context about both navigation and response at the same time. The result is more complete, accurate replies and actions.
Starting this week, you can opt in early to begin testing the new behavior. Navigate to your Gladly Agents → Basics → Early Access section → Unified Response Generation. We highly suggest turning this on now to validate your AI set up ahead of the full rollout.
A few tips to get the most out of this transition:
Review a sample of Sessions in the Conversation Review panel after enabling to confirm responses and navigation decisions meet your expectations.
Compare handoff rates and resolution rates before and after to measure the impact.
Reach out to Gladly Support with any questions or feedback.
Self-Healing Responses: AI Gets a Second Chance Before Handing Off
One of the biggest drivers of avoidable handoffs is a failed quality check – even when AI is one step away from a compliant answer. Self-Healing Responses gives the AI a second chance before handing off to an Agent.
Here's how it works:
When the proposed Gladly AI response fails a quality check, AI receives feedback from the failed check and regenerates the reply.
The new response is run through the full suite of quality checks again.
If the retry passes, the response is sent to the Customer.
If it still fails, the Conversation hands off to an Agent.
Your quality standards are not changing – every response must pass before it reaches the Customer. The difference is that more recoverable Conversations stay with AI instead of escalating on the first failed attempt.
Self-Healing Responses is now enabled by default across your Gladly AI set up. Note that this is not currently available for Voice AI.
If you still prefer to wait, you can opt out by navigating to your Gladly Agent → Basics → Early Access and clicking the Self-Healing Responses toggle to turn it off. Here is the rollout timeline:
Optional opt-out period, July 8–July 23: Self-Healing Responses is on by default. Click the toggle on any Gladly Agent to opt out during this window.
July 23: The opt-out toggle will be removed from the Basics tile and Self-Healing Responses will be the standard across Gladly Agents.
A few tips to get the most out of this transition:
Monitor your quality check pass rates and handoff volume after enabling to see how many Conversations are being recovered.
Review flagged Sessions in the Conversation Review panel to confirm retried responses meet your expectations.
Reach out to Gladly Support with any questions or feedback.
"Conversations" Renamed to "Sessions" in Gladly AI
Next week, we will rename "Conversations" to "Sessions" throughout the Gladly AI interface – most visibly on the "Gladly AI Conversations" page, which will become "Gladly AI Sessions."
The update reflects how these terms are already defined in the Gladly platform:
A Conversation is the time-bounded grouping of all interactions a Customer has with your brand. It can include any combination of AI Sessions and team-handled Contacts.
A Session is a single Gladly AI interaction – one Chat, one phone call, one Email thread handled by AI. It stays open while the Customer and Gladly AI are actively exchanging messages and closes after 2 hours of inactivity.
Because Gladly AI operates at the Session level – handling one interaction at a time – we’ll be updating the UI to clearly reflect this.
This is a terminology change only; reporting and billing are not affected. Learn more about Conversations vs. Sessions – and which reports to use for each.
Refreshed Main Navigation
Next week, we'll be updating the main navigation menu to make Gladly easier to navigate as it continues to grow.
The new menu organizes features into clear, intuitive categories – separating AI items and Gladly Team into their own sections, and grouping items by what you do with them (Analyze, Manage, Test, Deploy) rather than a flat list.
A few things to keep in mind:
Admin Settings stay the same – Existing Settings menu and selections are not affected.
Consistent across all pages – The full navigation menu will appear everywhere in the product, not just on certain pages.
Permissions are unchanged – What you see will continue to depend on your role and access level.
[Early Access] Shopify Product Search for Gladly AI
AI can search your Shopify catalog directly so product questions get answered with up-to-date store data.
Real-time catalog data – Product cards with the real image, current price, working link, and true variant and inventory availability, pulled from Shopify on each request.
Semantic, typo-tolerant search – Customers can ask naturally ("Do you have this in size 8?") and Gladly AI matches them to the right products.
Custom fields included – Details that live only in your custom fields – care instructions, dimensions, ingredient lists – are surfaced to Gladly AI so it can answer questions web search can't.
Web search stays in its lane – Product questions use the Shopify catalog; policies, FAQs, and blog content still come through web search.
Shopify Product Search is not a separate add-on – it's a new knowledge source for Gladly AI alongside web search and Answers. If you’d like to test this feature out early, reach out to Gladly Support to get started.
[Action Required] Contact Summary and Work Sessions Reports: Older Versions Deprecated Soon
We've migrated the Contact Summary, and Work Sessions reports to new infrastructure that makes them 2-3x faster!
As part of this migration, we will retire older versions of these reports. The following Contact Summary and Work Sessions report versions will be deprecated on September 2, 2026:
ContactSummaryReport
WorkSessionEventsReport
WorkSessionEventsReportV2
WorkSessionEventsReportV3
Chat Widget Accessibility
Fixed – Screen readers no longer announce the minimized Chat button as being inside a dialog after the Chat is closed, and keyboard focus no longer gets stuck on the button.
Empty Guide Section Instructions Restored
Fixed – For existing Guides whose instructions had been cleared out, guidance the AI follows has been restored for affected sections without any re-authoring on your part.
Stuck Conversations in the Automation Inbox
Fixed – Gladly AI Sessions could remain stuck open in the automation Inbox after an error while closing. They now close normally.
Complete Chat Handoff Transcripts on Retry
Fixed – If a Chat handoff's transcript was long enough to span multiple messages, or included attachments, and an error interrupted posting it, a retry could leave the Agent with a truncated transcript and missing images. Retries now deliver the complete transcript and all attachments.
Built-In Action Duplication in App Actions Builder
Fixed – Saving a built-in action from the App Actions builder could create a duplicate, misconfigured copy of that action. Built-in actions are now protected from accidental duplication.
Handoff Summary Visible in Conversation Review Panel
Fixed – The handoff summary could be missing when reviewing handed-off Conversations. Reviewers now consistently see the handoff summary and can click it to jump to the moment of handoff in the Inspect Summary.
Conversation Review Panel Shows All Gladly Agents Involved
Fixed – The Inspect Summary shows the full set of Gladly Agents involved within an AI Session. Previously the Conversation Review panel showed only the Gladly Agent that the Conversation ended under, with no indication it had started elsewhere or been re-routed.
Custom Channel Replies for Multi-Device Customers
Fixed – Replies on a custom channel could fail to deliver for Customers who had messaged from more than one device. The Composer now defaults the reply recipient to the device the Customer most recently messaged from, so Agents no longer have to manually switch recipients.
Gladly AI for Helpdesks Fixes
Fixed – A duplicate handoff for the same Chat could send the Customer a "Something went wrong" message and email an error to the connected helpdesk even though the handoff was successful. Duplicate deliveries are now handled silently.
Fixed – A Chat handoff whose Topic contained a space or comma could create several unintended tags on a Zendesk ticket instead of one.
Fixed – A Chat message with no text and no attachments could stop later messages in that Conversation from syncing between Gladly and the connected helpdesk.
Fixed – Transcripts for AI-resolved Chat messages now include the AI's product recommendations instead of dropping them.
RESOURCE
ON-DEMAND: July 2026 Gladly Roadmap Webinar
Want a preview of the latest innovations from the Gladly Product and Engineering team?
Watch this on-demand webinar to explore recent product releases, get a look at what's coming next, and see the features designed to help your team work smarter and deliver even better customer experiences.
REVIEWS
You're In Good Company
Love using Gladly? Share your experience on G2 Crowd to help others discover how Gladly empowers better customer connections.
REVIEWS
Using Gladly with your Shopify store?
Using Gladly with your Shopify store? We'd love your feedback there too! Leave us a review on the Shopify App Store and help other merchants discover how Gladly creates seamless, revenue-driving conversations.
RESOURCE
Gladly Referral Promotion
Love using Gladly? Now’s your chance to share the experience with others in your professional network. Submit your B2C or e-commerce referral to receive up to $2,000! Visit the Refer-A-Friend program to learn how.
July 8, 2026
Self-Healing Responses: AI Gets a Second Chance Before Handing Off
One of the biggest drivers of avoidable handoffs is a failed quality check – even when AI is one step away from a compliant answer. Self-Healing Responses gives the AI a second chance before handing off to an Agent.
Here's how it works:
When the proposed Gladly AI response fails a quality check, AI receives feedback and regenerates the reply.
The new response is run through the full suite of quality checks again.
If the retry passes, the response is sent to the Customer.
If it still fails, the Conversation hands off to an Agent.
Your quality standards are not changing – every response must pass before it reaches the Customer. The difference is that more recoverable Conversations stay with AI instead of escalating on the first failed attempt.
Starting this week, Self-Healing Responses will be enabled by default across your Gladly AI set up. Note that this is not currently available for Voice AI.
If you prefer to wait, you can opt out by navigating to your Gladly Agent → Basics → Early Access and clicking the Self-Healing Responses toggle to turn it off.
Here is the rollout timeline:
Optional opt-out period, July 8–July 23: Self-Healing Responses is on by default. Click the toggle on any Gladly Agent to opt out during this window.
July 23: The opt-out toggle will be removed from the Basics tile and Self-Healing Responses will be the standard across Gladly Agents.
A few tips to get the most out of this transition:
Monitor your quality check pass rates and handoff volume after enabling to see how many Conversations are being recovered.
Review flagged Conversations in the Conversation Reviewer to confirm retried responses meet your expectations.
Reach out to Gladly Support with any questions or feedback.
Workflow Authoring Improvements
Five updates to the Workflows editor make the build-out process more intuitive and effective:
Clear condition operators – The editor labels each "Contains" operator in plain English ("Contains the text," "Contains any of these," "Contains any of these whole words") and displays list-valued operators as removable chips instead of a single comma-separated text box, making it obvious whether a condition takes one value or a list.
‘Stop’ steps can target a specific Guide section – ‘Stop’ steps that route to a Guide can now target a specific section instead of always starting from the beginning, so a Workflow can send the Customer straight to the relevant part.
Required-input validation for nested Workflows – The editor flags any step that includes a nested Workflow without providing one of that Workflow's required inputs.
Field-aware ‘Ask’ step questions – AI asks more specific opening questions when an ‘Ask’ step collects information. Previously the first question could be generic, it now reflects the actual fields being collected, so Customers get a clearer prompt.
Apply Freeform Topics – The Apply Freeform Topics action is available within the Workflows editor. Use it to set one or more Freeform Topic on the current Conversation as part of a Workflow run.
Clickable Links in Historic Guide Version
When reviewing a previous version of a Guide, handoff and Rule links inside a section are now clickable – reviewers can open the link to see the handoff reason or the Rule's conditions. Previously these links were inactive when viewing a historical version.
Granular Settings Permissions for Custom Roles
Admins can grant Custom Roles access to the following individual Settings areas via four new permissions:
Agent Experience
Channels
Orchestration
Proactive Conversations
Absolute Timestamps in the Conversation Timeline
Agents can enable absolute timestamps to display for each message in the Conversation Timeline.
Go to Profile → Detailed Dates to switch from relative timestamps ("2 hours ago") to full date-and-time stamps. Messages from a previous year also include the year.
Custom Notification Sounds and Channel Colors
Agents can personalize their Gladly Team experience with per-Channel notification sounds and Channel Contact border colors – both configurable from Profile settings.
Assign a distinct sound to each Channel so you can tell a Chat from an SMS or Email by ear, and assign a distinct border color to each Channel Contact on the Conversation Timeline so you can easily tell them apart at a glance.
Shopify App v5: More Actions Gladly AI Can Handle Directly
Shopify App v5 expands what Gladly AI can do on the Customer's behalf, covering more of the order lifecycle end-to-end without handing off to an Agent.
Replacement orders and draft orders – AI can build an order on the Customer's behalf and let them pay through Shopify. Create a new draft order and email the Customer a Shopify checkout link to pay or complete the draft into an order directly.
Deeper order visibility – A new ‘orderByIdentifier’ action gives Gladly AI the complete picture of a single order: every line item's lifecycle (ordered, shipped, refundable, or removed), shipments with tracking and delivery estimates, returns, refunds, and fulfillment holds.
Cancel and refund – Cancel an order and refund to the original payment method or to store credit, with control over restocking and Customer notification.
One-time discount codes – Issue a single-use, single-Customer discount code scoped to the whole order or to specific collections, products, or variants.
Live inventory checks – Check product inventory before promising a replacement or reorder, so Customers get real-time answers.
Invoice emails – Send an invoice email for an unpaid order.
How to upgrade: Shopify v5 requires a one-time re-authorization and a review of any Guides or Rules that reference changed actions.
Re-authorize the app. v5 requests additional Shopify permissions. Sign in to Shopify from the app's configuration and approve the updated permissions – the new actions will not work until you do.
Update Guides that reference changed actions. A few actions were rebuilt in v5, and a Guide configured against the old shapes needs a one-time update. Be sure to review any Guide that uses the following actions, and revise them as necessary:
lookupOrderByOrderNumlookupOrdersByEmailsuggestedRefundcancelOrderupdateShippingAddress
Review Rules against the streamlined order data. Order lookups return a leaner shape – each line item carries a compact product and variant reference, and rarely-used fields (transactions, billing address, fulfillment details) have moved to
orderByIdentifier. If a Rule references a removed field, point it at the new shape or fetch the detail throughorderByIdentifier.
[Action Required] Contact Export, Contact Summary, and Work Sessions Reports: Older Versions Deprecated Soon
We've migrated the Contact Export, the Contact Summary, and Work Sessions reports to new infrastructure that makes them 2-3x faster!
As part of this migration, we will retire older versions of all three of these reports, with ContactExportReport and ContactExportReportV2 deprecated first, next week. If your team currently uses either of these versions, please update your integrations to use ContactExportReportV3 to avoid any disruption.
The following Contact Summary and Work Sessions report versions will be deprecated on September 2, 2026:
ContactSummaryReport
WorkSessionEventsReport
WorkSessionEventsReportV2
WorkSessionEventsReportV3
"Conversations" Renamed to "Sessions" in Gladly AI
Next week, we will rename "Conversations" to "Sessions" throughout the Gladly AI interface – most visibly on the "Gladly AI Conversations" page, which will become "Gladly AI Sessions."
The update reflects how these terms are already defined in the Gladly platform:
A Conversation is the time-bounded grouping of all interactions a Customer has with your brand. It can include any combination of AI Sessions and team-handled Contacts.
A Session is a single Gladly AI interaction – one Chat, one phone call, one Email thread handled by AI. It stays open while the Customer and Gladly AI are actively exchanging messages and closes after 2 hours of inactivity.
Because Gladly AI operates at the Session level – handling one interaction at a time – we’ll be updating the UI to clearly reflect this.
This is a terminology change only; reporting and billing are not affected. Learn more about Conversations vs. Sessions – and which reports to use for each.
Refreshed Main Navigation
We'll be updating the main navigation menu the week of July 22, to make Gladly easier to navigate as it continues to grow.
The new menu organizes features into clear, intuitive categories – separating AI items and Gladly Team into their own sections, and grouping items by what you do with them (Analyze, Manage, Test, Deploy) rather than a flat list.
A few things to keep in mind:
Admin Settings stay the same – Existing Settings menu and selections are not affected.
Consistent across all pages – The full navigation menu will appear everywhere in the product, not just on certain pages.
Permissions are unchanged – What you see will continue to depend on your role and access level.
Signals: AI Conversation Quality Evaluation
Signals is a new way to evaluate the quality of your AI Conversations at scale – no surveys, manual review, or third-party tools required. Write evaluation criteria in plain English, run them against hundreds of Conversations at once, and see pass/fail breakdowns with the ability to drill into any result for the full transcript and AI reasoning.
Signals gives you continuous visibility into how your AI is performing, at a scale no team could match manually. Define evaluation criteria around the things that matter most to your operation – sentiment, policy adherence, customer feedback on a new product change, or anything else you want to track.
Write your criteria in plain English, run them against hundreds of Conversations at once, and get qualitative insights with the ability to drill into any result for the full transcript and AI reasoning. Monitor every AI Conversation and uncover Customer insights, right in Gladly.
Want early access? Signals is currently in beta and will be available for free temporarily. If you'd like to try it and share your feedback, please contact Gladly Support to get started.
Voice Mode Coming to the Guide Tester
Soon you'll be able to test how your Voice AI sounds to a caller before they experience it – without picking up the phone.
Voice testing is coming directly to the Guide Tester. Instead of typing, click the microphone icon, speak naturally, and hear the AI respond out loud using the same speech recognition, spoken voice, and pronunciation settings as a real phone call. See exactly how your Guide performs in a real voice conversation – the response timing, the language, the tone, all of it.
It's a private test conversation that never reaches live Customers or affects your reporting, so you can validate how callers will be understood and how responses will sound before going live. The full transcript appears in the tester afterward for review. No phone number or IVR setup is required – just open the Guide Tester, select Voice as the Channel, and start talking.
Before this, validating Voice AI behavior meant placing actual phone calls. Now you can validate the full voice experience right inside Gladly, catch issues before they reach callers, and tune your setup with confidence. Stay tuned for availability details.
Multiple Channels Supported in Workflow Conditions
Fixed – A Workflow Channel condition using "Contains any of these" can now match multiple Channels. Previously the selector was limited to one.
App Platform Actions With Long Descriptions Import Correctly
Fixed – An organization's Shopify (App Platform) actions could fail to import or appear in the Guide editor when any action had a description over 1,024 characters. Long descriptions are now handled correctly.
Workflow ‘Ask’ Step No Longer Fails on First Message
Fixed – A Workflow that opens by asking the Customer a question could fail when triggered at the very start of a Session (before the Customer had sent any message). These Workflows now reliably produce their opening question.
Add Note Preserves Line Breaks in Gladly Team
Fixed – The Add Note action now preserves authored line breaks. Additionally, using Enter in the Note field now inserts a line break (Shift+Enter still works).
Markdown Links No Longer Incorrectly Flagged as Unverified
Fixed – A valid link could be incorrectly flagged as unverified when AI formatted it as a Markdown link whose visible text was the URL itself. These types of links are now validated correctly.
Conversations Open in a New Tab
Fixed – Conversations on the Gladly AI Conversations page can now be opened in a new browser tab. Right-click a Conversation to open it in a new tab.
Agent Liveboard Fixes
Fixed – Hovering over "+ N more" in an Agent's "Currently On" column on the Agents Liveboard correctly displays the popup listing that Agent's other active Sessions.
Fixed – An Agent on a phone call no longer disappears from the Agents Liveboard's "Currently On" view when they mute themselves or put the Customer on hold.
Fixed – A live phone call an Agent was actively handling could show a "No agent reply" indicator on the Agents Liveboard. Phone calls are correctly marked as active.
Category-Admin Roles Can Access All Settings Page Dependencies
Fixed – Admins whose role grants only a single Settings category permission (Channels, Orchestration, Agent Experience, or Proactive) could be bounced to the home page or see missing data when opening pages in that category.
This affected creating or editing Chat configurations, setting a Chat logo, and opening the Proactive Voice campaign editor. All category-specific Settings pages now load correctly for the corresponding role.
Customer Deletion for Large Conversation Histories
Fixed – Customer profiles with very large Conversation histories could not be deleted by the data-retention job or compliance deletion. These deletions now complete successfully.
App Platform Errors Are More Descriptive
Fixed – Errors from App Platform apps are clearer and more actionable. This applies everywhere errors are surfaced:
App Platform calls in the Conversation Reviewer
App Platform Action and Custom Action logs in the Gladly UI
‘appcfg’ CLI when testing, developing, or reviewing detailed logs for app usage
Previously, error messages could be vague. Resource-limit breaches (output size or execution time) now also report a dedicated error code.
RESOURCE
Who Should Own AI in a Customer Experience Team?
The strongest predictor of AI performance wasn't the model, the vendor, or the number of integrations built. It was whether one person was clearly accountable for results.
But who should own AI in CX?
RESOURCE
How Do You Know if Your AI is Actually Working?
Most AI teams obsess over deflection rate. How many conversations did the AI touch? How many stayed away from a human?
But when Gladly studied what retail's best AI deployments have in common, a different pattern emerged.
REVIEWS
You're In Good Company
Love using Gladly? Share your experience on G2 Crowd to help others discover how Gladly empowers better customer connections.
REVIEWS
Using Gladly with your Shopify store?
Using Gladly with your Shopify store? We'd love your feedback there too! Leave us a review on the Shopify App Store and help other merchants discover how Gladly creates seamless, revenue-driving conversations.
RESOURCE
Gladly Referral Promotion
Love using Gladly? Now’s your chance to share the experience with others in your professional network. Submit your B2C or e-commerce referral to receive up to $2,000! Visit the Refer-A-Friend program to learn how.
July 1, 2026
Gladly Agent Versions and Entry Points Updates
A batch of improvements to the Versions page, Entry Points page, and Guide tester make it easier to test, deploy, and iterate on your Gladly Agents.
Inspect Summary links to the specific Gladly Agent version – The Inspect Summary in the Conversation Review panel navigates directly to the version of the Gladly Agent that ran during the Conversation. If the Conversation ran on an older version, clicking the link opens a read-only view of that version. This also applies to test Conversations run in the Guide tester.
Select a Gladly Agent version in the Guide tester – Choose which Gladly Agent version to use when testing scenarios within the Guide tester.
Search bar on Add Entry Points – The "Add entry points" button on the Entry Points page includes a search bar, making it easier to find unassigned Entry Points by name or address instead of scrolling the full grouped list.
Workflow Trigger changes tracked in version history – Changes made to the Workflow Triggers tile on the Gladly Agent page are now reflected in the "What Changed" column on the Versions page.
"Workflows" label renamed to "Workflow Triggers" on Versions page – The "Workflows" label displayed in the "What Changed" column has been renamed to "Workflow Triggers" for clarity.
Refreshed Gladly AI Conversations Page
The Gladly AI Conversations page has an updated look and feel. Each row displays the first message sent by the Customer and the first response generated by AI, making it easier to scan and identify sessions at a glance. The Channel icon has also been relocated to the start of each row for quicker visual reference.
Granular Settings Permissions for Custom Roles
Admins can grant Custom Roles access to the following individual Settings areas via four new permissions:
Agent Experience
Channels
Orchestration
Proactive Conversations
Dedicated Hero Condition in Workflow ‘If’ Step
Build ‘If’ step conditions on a Customer's Dedicated Hero using the new ‘Dedicated Hero’ variable. Check whether the Customer has a Dedicated Hero, or matching on the assigned Agent's name, email, or team name – instead of writing a custom expression.
Inline Workflow Step Review in the Tester
Workflow authors can review all the steps a tested Workflow took directly within the Workflow tester. Expand each message the Workflow sent for a deeper view of how it arrived at the response and debug behavior without leaving the Workflow editor.
Gladly AI Sessions Reporting Now Includes Historical Data
Gladly AI Sessions reporting includes historical data from before January 2026. Previously, Sessions reporting was limited to data on or after January 1, 2026, which meant teams couldn't compare current AI performance against earlier periods or analyze long-term trends. The full history is now available, making it easier to measure progress, identify seasonal patterns, and build reports that span your entire Gladly AI rollout.
Updated AI performance: All Gladly Agented rolled over to new models
As first shared in the June 10 release notes, Gladly has been migrating all organizations to updated AI models to ensure Customers benefit from the latest advances in performance and accuracy.
Updated models are now the default for all Gladly Agents, and the "Updated AI performance" toggle has been removed from the 'Basics' tile.
Phone Metrics Added to Insight Builder
New phone-related metric families are now available in the Agent Metrics explore in Insight Builder:
Contact Handle Time – Phone Call and After Contact Time – Phone Call, broken out by Inbound and Outbound.
Contacts With Contact Handle Time, broken out by Inbound/Outbound for Phone Call, Chat, SMS, and Email.
Hold Time – Phone and Talk Time – Phone, broken out by Inbound and Outbound (previously available only at the Contact level).
New App Platform Integrations: Klaviyo and Okendo
Two additional external apps are now available on the App Platform, bringing email marketing and reviews-and-loyalty data directly into the Conversation:
Klaviyo – Surfaces a Customer's email marketing consent status, recent marketing emails, and subscription profile in a Customer Details card.
Agents can see what a Customer has received before replying, and unsubscribe a Customer from Klaviyo marketing email directly from the Conversation, no tool-switching required.
Klaviyo data is also available to Gladly AI via Data Pulls, so Guides can factor in subscription status when responding.
Okendo – Surfaces a Customer's product reviews and loyalty standing in two Customer Details cards:
Okendo Reviews (star rating, content, photos, sentiment, and any store reply)
Okendo Loyalty (points balance, enrollment status, and VIP tier)
Agents can reply to a review – publicly or privately – without leaving Gladly
Okendo loyalty data is available to Gladly AI via Data Pulls for personalized, context-aware responses.
To get started, navigate to Settings → Apps → Create Configuration and select Klaviyo or Okendo from the menu.
[Action Required] Contact Export, Contact Summary, and Work Sessions Reports: Older Versions Deprecated Soon
We've migrated the Contact Export, the Contact Summary, and Work Sessions reports to new infrastructure that makes them 2-3x faster!
As part of this migration, we will retire older versions of all three of these reports, with ContactExportReport and ContactExportReportV2 deprecated first, in 2 weeks. If your team currently uses either of these versions, please update your integrations to use ContactExportReportV3 to avoid any disruption.
The following Contact Summary and Work Sessions report versions will be deprecated on September 2, 2026:
ContactSummaryReport
WorkSessionEventsReport
WorkSessionEventsReportV2
WorkSessionEventsReportV3
"Conversations" Renamed to "Sessions" in Gladly AI
In two weeks, we will rename "Conversations" to "Sessions" throughout the Gladly AI interface – most visibly on the "Gladly AI Conversations" page, which will become "Gladly AI Sessions."
The update reflects how these terms are already defined in the Gladly platform:
A Conversation is the time-bounded grouping of all interactions a Customer has with your brand. It can include any combination of AI Sessions and team-handled Contacts.
A Session is a single Gladly AI interaction – one Chat, one phone call, one Email thread handled by AI. It stays open while the Customer and Gladly AI are actively exchanging messages and closes after 2 hours of inactivity.
Because Gladly AI operates at the Session level – handling one interaction at a time – we’ll be updating the UI to clearly reflect this.
This is a terminology change only; reporting and billing are not affected. Learn more about Conversations vs. Sessions – and which reports to use for each.
Refreshed Main Navigation
We'll be updating the main navigation menu to make Gladly easier to navigate as it continues to grow.
The new menu organizes features into clear, intuitive categories – separating AI items and Gladly Team into their own sections, and grouping items by what you do with them (Analyze, Manage, Test, Deploy) rather than a flat list.
A few things to keep in mind:
Admin Settings stay the same – Existing Settings menu and selections are not affected.
Consistent across all pages – The full navigation menu will appear everywhere in the product, not just on certain pages.
Permissions are unchanged – What you see will continue to depend on your role and access level.
Stay tuned for more details in the coming weeks.
Signals: AI Conversation Quality Evaluation
Signals is a new way to evaluate the quality of your AI Conversations at scale – no surveys, manual review, or third-party tools required. Write evaluation criteria in plain English, run them against hundreds of Conversations at once, and see pass/fail breakdowns with the ability to drill into any result for the full transcript and AI reasoning.
Signals gives you continuous visibility into how your AI is performing, at a scale no team could match manually. Define evaluation criteria around the things that matter most to your operation – sentiment, policy adherence, customer feedback on a new product change, or anything else you want to track.
Write your criteria in plain English, run them against hundreds of Conversations at once, and get qualitative insights with the ability to drill into any result for the full transcript and AI reasoning. Monitor every AI Conversation and uncover Customer insights, right in Gladly.
Want early access? Signals is currently in beta and will be available for free temporarily. If you'd like to try it and share your feedback, please contact Gladly Support to get started.
Voice Mode Coming to the Guide Tester
Soon you'll be able to test how your Voice AI sounds to a caller before they experience it – without picking up the phone.
Voice testing is coming directly to the Guide Tester. Instead of typing, click the microphone icon, speak naturally, and hear the AI respond out loud using the same speech recognition, spoken voice, and pronunciation settings as a real phone call. See exactly how your Guide performs in a real voice conversation – the response timing, the language, the tone, all of it.
It's a private test conversation that never reaches live Customers or affects your reporting, so you can validate how callers will be understood and how responses will sound before going live. The full transcript appears in the tester afterward for review. No phone number or IVR setup is required – just open the Guide Tester, select Voice as the Channel, and start talking.
Before this, validating Voice AI behavior meant placing actual phone calls. Now you can validate the full voice experience right inside Gladly, catch issues before they reach callers, and tune your setup with confidence. Stay tuned for availability details.
We're Bringing Mobile Support to Gladly!
We are excited to introduce mobile support for Gladly Team. Designed for Agents on the go or away from their desk, this update lets you view your Customer list, open Conversations, and take action directly from your phone.
Want early access? Mobile support is currently in beta. If you'd like to be among the first to try it and share your feedback, please contact Gladly Support to get started.
Versions Page Timezone Incorrect on Refresh
Fixed – Timestamps in the Version and Summary columns on the Versions page would briefly render in the wrong timezone on page refresh before correcting to the local time.
Inspect Summary Shows Local Timezone
Fixed – The Agent version timestamp in the Inspect Summary panel now displays in the local timezone instead of UTC.
Rich-Content Agent Messages Trigger message:received Event
Fixed – The message:received event now fires for Agent messages sent as rich (HTML) content. Previously these messages did not trigger the event.
Workflow Editor, Runtime, and Template Fixes
Fixed – Using the same action more than once in a Workflow produced a duplicate-output error that prevented the author from saving. Repeated actions now save correctly, and all validation errors are listed under the validation banner with a working "Jump to" link that navigates to the exact field.
Fixed – The error banner in the Workflow editor could remain stuck on "Save blocked until errors are resolved,” with no visible per-step errors.The banner now reflects the errors actually shown on the steps, so resolving them clears the banner and re-enables Save.
Fixed – Several Workflow templates had starter conditions that referenced variables incorrectly, appearing as unrecognized custom conditions or silently failing to evaluate.
Deployment Status Banner Incorrectly Showing "Not Published"
Fixed – The deployment status banner on the Gladly Agent page could incorrectly show "Not published" even when the Gladly Agent was actively assigned to Entry Points.
Deployment Status Banner Incorrectly Showing "Published"
Fixed – The deployment status banner on the Gladly Agent page could incorrectly show "Published" even when no Entry Points were connected to the Gladly Agent.
Version Preview for Entry Points Linked to Older Versions
Fixed – The version preview stopped working and the label showing how many iterations a version was behind could disappear for Entry Points linked to versions more than 25 saves old.
Shared Report Editing Restricted to Analysts
Fixed – Editing and deleting shared Insight Builder reports from the Reporting view is now restricted to users with the Analyst role. Other users with report access retain view-only permissions.
Analyst Role Includes Read Reports Permission
Fixed – The built-in Analyst role includes the "Read reports" permission, allowing users with only the Analyst role to load Insight Builder and reports without needing to be combined with another role.
OrderGroove Product Names Display for Large Subscriptions
Fixed – Product names in the OrderGroove app were not displaying when a Customer had more than ten open subscriptions.
EVENT
NEXT WEEK! Gladly Roadmap Webinar
Join us on Thursday, July 9 at 11 a.m. PT for a preview of the latest innovations from the Gladly Product and Engineering teams.
We'll cover recent releases, share what's coming next, and give you an early look at the features designed to help your team work smarter and serve customers better.
RESOURCE
5 AI Prompts CX Leaders Actually Need
Five prompts to help you find what’s quietly broken in your CX operation and make the case for fixing it in a language your CFO actually speaks.
REVIEWS
You're In Good Company
Love using Gladly? Share your experience on G2 Crowd to help others discover how Gladly empowers better customer connections.
REVIEWS
Using Gladly with your Shopify store?
Using Gladly with your Shopify store? We'd love your feedback there too! Leave us a review on the Shopify App Store and help other merchants discover how Gladly creates seamless, revenue-driving conversations.
RESOURCE
Gladly Referral Promotion
Love using Gladly? Now’s your chance to share the experience with others in your professional network. Submit your B2C or e-commerce referral to receive up to $2,000! Visit the Refer-A-Friend program to learn how.
.svg?sv=2026-02-06&spr=https&st=2026-07-22T00%3A37%3A22Z&se=2026-07-22T01%3A22%3A22Z&sr=c&sp=r&sig=tJUfRz8l97sDWYrQEA4PjE0rv884IJOhh2NIbTzDR%2BA%3D)
on the top-left corner, then click Contact Gladly Support. Gladly Support's primary operating hours are Monday through Friday, 6 AM to 6 PM Pacific Time, but you can submit a request anytime.