the voice of customer experience technology
Front page
NewsService Operations1h · 15:01 BST · 12 min read

Grafana's Grades, Contact Center Handoffs and the Carrier Test

At ObservabilityCON, Grafana introduced new investigation tools and previewed ways to link user experience with service ownership. Zoom and Genesys are tightening contact center handoffs, while fresh carrier incidents test whether suppliers can tell when customers actually get through.

Service management observability

There’s a special kind of confidence you only get from a conference demonstration. An AI agent makes the right decision, the dashboard behaves itself, and everybody goes home excited. Try adding a missed on-call alert or a carrier that won’t connect the customer’s call, and things get rather more complicated.

Last week, we handed Grafana five questions ahead of ObservabilityCON. This week, we’re marking the answers. Its AI observability usage charges began October 1, so customers have every reason to ask what that extra visibility buys them when something actually goes wrong.

Zoomtopia has a rather different test. Zoom’s now measuring whether customers return after an AI interaction, and its October contact center release puts more useful context in front of human agents. Genesys is bringing more ServiceNow work into the same queue, with live voice transcription confirmed in its October 19 release. I’d like to know what happens when either handoff loses the conversation or its owner.

ServiceNow’s London forum offers another chance to see autonomous service operations in practice, while Freshworks has widened the asset and dependency information available to IT teams. Both approaches could help when a customer-facing service breaks. The question is how quickly.

We’ve heard plenty of plans. Here’s where the vendors are making progress, what their evidence proves, and which questions still need an answer.

TL;DR

  • Grafana’s ObservabilityCON report card now has confirmed releases to assess: generally available Assistant Watchers and OpenTelemetry-enabled Agent Observability, plus expanded Knowledge Graph context. Customer-journey and incident-response additions remain in preview, leaving questions about costs, human approval, and service recovery.

  • Genesys’s October 12 ServiceNow updates improve email routing and make accepted work easier to measure. Its October 19 release also adds real-time voice transcripts to ServiceNow Agent Workspace, subject to eligible Genesys Cloud licensing and regional rollout.

  • Zoom’s October 17 Contact Center release improves ServiceNow screen pops and call routing. Zoomtopia’s new Unified Context Layer aims to preserve customer history across interactions, while U.S. Oral Surgery Management uses Zoom Contact Center and Phone data to follow patient transfers between centralized scheduling and individual practices.

  • Freshworks has expanded Freshservice’s asset visibility, while Grafana’s October 19 paging update and Twilio’s new Brazil and Thailand voice incidents show why incident recovery needs careful measurement. Infobip’s October 19–20 messaging problems add another supplier to that test.

What Did Grafana Announce at ObservabilityCON 2026?

Grafana picked an awkward month to make its case. AI observability charges began October 1, while its notification history raised uncomfortable reliability questions. ObservabilityCON supplied some actual releases to judge: Assistant Watchers and OpenTelemetry-enabled Agent Observability are generally available. Watchers can catch changes missed by ordinary alerts, and Knowledge Graph now spans Grafana Cloud stacks. Service ownership mapping remains in private preview.

There’s a customer angle, too. Session Replay is in public preview and Critical User Journeys in private preview. They could help explain a failed web interaction; I’d still like Grafana to trace a dropped contact center transfer. LexisNexis Risk Solutions described an investigation-to-ServiceNow-to-on-call workflow. Its Neil Wilson highlighted “connecting service ownership, dependencies, operational knowledge, and customer impact.” Who checked recovery?

Then there’s the bill. Faster full-text log search is generally available without an additional charge. Grafana Pro still lists $1.50 per 1,000 additional agent generations and $2 per million evaluation or guard tokens after allowances, with other platform, telemetry, and model costs separate. Price one resolved incident for us.

Grafana’s notification history complicates the reliability pitch. On October 19, it closed the Thailand SMS case by clarifying that some destinations aren’t supported. Did backup push or phone alerts reach the engineer? That’s a rather important distinction from “recovered.”

Managed Sandboxes, in public preview, can prepare proposed fixes for human approval. Grafana Workflows is also in public preview; the October 21 IRM session promises people control over findings. If a suggested fix changes contact center routing, who signs off and who can reverse it?

What Did Genesys Change About Contact Center and ITSM Handoffs?

Genesys and ServiceNow have made a series of practical improvements to service management and contact center handoffs over the past few months.

On October 12, Genesys updated Unified Experience so ServiceNow email details, including the sender, subject, and message content, can help route work by priority or skill. Eligible emails in ServiceNow CSM Workspace can also receive automatic acknowledgments. Handy, although I wouldn’t count a polite auto-reply as a resolved customer problem.

The integration already routes ServiceNow ITSM incidents as work items, while September brought agent-first callbacks and July expanded ITSM digital messaging. That could save customers unnecessary transfers, provided somebody actually accepts the work when it reaches their queue.

Genesys made that distinction easier to measure on October 12. The old ‘Answered’ metric included work offers agents declined or let time out. Those now count as ‘Answer Attempted’, while ‘Answered’ measures accepted work. It’s a welcome correction, and I’d happily see other vendors make it too. The change applies to the Work Automation add-on.

Live voice has now arrived, too. Genesys’s October 19 release confirms real-time transcripts in ServiceNow Agent Workspace, where Now Assist can use them during calls. That could spare agents plenty of retyping. Eligible licenses include CX 1 with Digital Add-on II and CX 2, 3, or 4 variants, with regional rollout scheduled through the week.

What Did Zoom Announce at Zoomtopia 2026 for Contact Centers?

Zoomtopia gave us something more useful than another containment percentage. Zoom’s new Unified Context Layer aims to preserve customer history across Contact Center, Phone, and other interactions. U.S. Oral Surgery Management already uses Zoom Contact Center and Phone data to follow patient calls from centralized scheduling to individual practices across more than 130 locations. That’s closer to the handoff evidence I’d like to see, although I’d still want to know what happens when a transfer fails.

September’s ‘resolved without return’ metric gets closer. Zoom counts a contained chat or voice interaction as resolved only if that customer doesn’t return within a defined window. I’d ask how long the window lasts and whether a follow-up call after a chat counts.

More interesting for service teams is Zoom’s October 17 Contact Center release, now marked ‘Released’. ServiceNow screen pops use a single window, while CTI panels can display Virtual Agent summaries. Administrators can group recommended transfer queues and adjust waiting callers’ priority. Zoom also lists outbound BYOC/BYOP number-based routing, subject to enablement. Those are useful changes to test on a real call. The planned warm-transfer-to-flow feature was pulled after deployment problems. Better that than discovering the problem mid-transfer.

There’s a deadline too. Customers using the legacy CTI V2 adapter URL, including ServiceNow integrations, should update it before December 1 to avoid potential service disruption. September’s analytics changes distinguish system-disconnected calls from customer hang-ups.

Zoom’s maintenance diary lists October 19–21 work moving Zoom Phone SIP registrations from AMS3 to FRA3, with possible brief disconnections. That isn’t a confirmed Contact Center outage. Still, I’d like Zoomtopia to show what happens when a transfer drops: who notices, and can the caller get through on the second try?

What Did ServiceNow Announce at World Forum London 2026?

ServiceNow heads into its October 20 London forum with plenty to explain. Its October 6 AI Workflow Factory is available, while Autonomous Engineer remains in early access. We’ve already covered the wider AI governance promises. I’d rather know what happens when a customer-facing workflow breaks and the service desk has to find somebody who can fix it.

Freshworks has an answer to part of that headache. Its October 15 Freshservice update adds standalone ITAM alongside existing ITSM tools, more data-center asset detail, and Cloud Collector discovery. Dependency mapping isn’t new, and its Discovery AI Agent remains in early access. But if half a contact center loses CRM access, could IT identify the failing dependency before agents start raising separate tickets?

Maybank signed a five-year ServiceNow agreement to expand its Resilience Operations Center across Malaysia, Singapore and Indonesia. It’s a useful service-dependency story, although I’d wait for measured recovery improvements before handing out praise.

ServiceNow’s financial-services Operational Resilience solution is available, while AI Passport, shadow-AI discovery and AI Cost Management are expected in Q4. Freshworks also agreed to acquire Torii, potentially extending SaaS and license visibility. The deal hasn’t closed, so those capabilities aren’t available yet.

Gartner’s October 19 research adds an uncomfortable comparison. Only 13% of CIOs reported significant or transformational AI value, while 51% expect technology-solution costs to increase because of AI. Those aren’t contact center figures, but I’d still like London to produce a named customer, a measured recovery time, and somebody willing to explain what actually improved.

Did the Carrier Layer Hold Up This Week?

I’d hate to spend a whole week judging intelligent service tools and forget whether customers could actually make a phone call.

Twilio’s October 19 board recorded voice-call failures and post-dial delays involving some Twilio mobile numbers calling Brazil. Separately, inbound calls to a subset of its Thailand numbers were failing. Different routes, different problems. A contact center trying to diagnose either needs to know which direction the call was traveling and which carrier handled it.

Thailand had another problem brewing. Post-dial delays affecting some calls to the country had persisted since October 12, with Twilio still working alongside network partners on October 19. I’d be asking whether backup numbers using the same downstream carrier helped anybody.

And Twilio wasn’t alone. Infobip reported a four-minute degradation in its EU region-locked environment on October 19, affecting SMS, email, RCS, voice, WhatsApp and Viber. A separate US SMPP disruption continued into October 20. That’s quite a collection of ways for customer communications to go wrong.

There’s another complication. Calls that never reach the IVR might never appear in the contact center’s queue reports. How many customers gave up before anyone noticed? Supplier status updates are useful, but I’d want independent test calls before declaring recovery.

Twilio also scheduled October 19–22 maintenance, including London interconnection work and messaging changes. Sinch had an October 20 SMS API certificate renewal planned. Neither is evidence of an outage, although both deserve attention from teams managing custom integrations.

A verification code arriving after it expires, or a callback that never connects, won’t look particularly successful to the customer waiting for it.

In Other News

Two smaller service-management updates caught my attention this week. On October 19, incident.io made Nexus Extensions generally available to Investigations customers, letting teams bring internal debugging tools and procedures into incident investigations. Access policies restrict what autonomous investigations can execute. Handy when a contact-center integration fails, and the evidence is scattered across Slack.

Genesys had a less exciting October 19 change: two older TLS cipher suites are no longer supported on BYOC Cloud trunks. Customers using only those suites need to add a supported option, while those already using another supported cipher needn’t change anything. Not much of a conference headline, unless calls stop connecting.

Next week, ICMI’s Contact Center Expo runs October 26–29 in Orlando. I’d like to hear from somebody who’s spent an afternoon chasing a failed carrier connection while three suppliers insisted their own systems were working.

Earnings should make for a different conversation. ServiceNow reports October 28, followed by Freshworks on November 3. We’ll cover those figures separately, although I’ll be watching for evidence that service operations are improving beyond the product demos.

Had a supplier mark an incident resolved while your contact center was still dealing with the fallout? I’d be interested to hear how you handled it.

Everyone Wants to Be the Control Layer. Who’s Earned It?

I’m not handing anyone the crown just yet, but there’s progress worth recognizing.

Grafana is making investigations more useful across telemetry and incident workflows. Zoom and Genesys are helping customer context travel between contact center channels and service workspaces. That’s encouraging, provided somebody can act when the handoff breaks.

Imagine Zoom’s virtual agent tells a customer their booking has changed, but the CRM never saves the update. Grafana spots the failure, and ServiceNow opens an incident. Good progress. Who tells the customer their booking still needs fixing?

I’d expect the IT incident owner to coordinate the fix and the contact center to take responsibility for updating affected customers. That needs agreeing before the incident, especially when several suppliers are involved, and everyone insists their own system is working.

I haven’t seen a single vendor demonstrate ownership of that entire process yet. There are useful improvements here, but I’d still want somebody checking that the customer’s booking was corrected before declaring the incident closed.

FAQs

What Did Grafana Announce at ObservabilityCON 2026?

Grafana announced generally available Assistant Watchers and OpenTelemetry-enabled Agent Observability, alongside expanded Knowledge Graph context. Critical User Journeys and several incident-response capabilities remain in preview. LexisNexis Risk Solutions described Grafana investigations connected to ServiceNow and on-call response, although verified customer recovery times remain the more demanding test.

How Does Genesys Connect Contact Center Work With ServiceNow?

Genesys Unified Experience sends ServiceNow CSM requests and ITSM incidents into Genesys queues. October’s updates improved email routing and made accepted-work reporting more accurate. The October 19 release also added live voice transcripts to ServiceNow Agent Workspace, but you’ll need to check licensing and regional availability before rolling it out.

What Were the Biggest Zoomtopia Contact Center Announcements?

Zoomtopia introduced a Unified Context Layer intended to preserve customer history across Contact Center and Phone. U.S. Oral Surgery Management uses data from both to follow transferred patient calls. Zoom’s October 17 release also improved ServiceNow screen pops and routing, although warm transfer into a voice flow was withdrawn after deployment problems.

Can Contact Center Calls Fail When the Provider Says Its Platform Is Working?

Absolutely. Twilio reported separate Brazil and Thailand voice problems on October 19, while Infobip recorded a brief disruption across several communication channels. A healthy CCaaS platform won’t help much if the carrier can’t connect the call. I’d test the affected routes before accepting an incident as resolved.

rate this story
helps rank stories across CX Today
The discussion0 takes · attributed & checked

Does this reflect your experience?

opening the room…
Read nextordered by techtelligence · every pick explained
same beat · Service Operations

PSTN Switch Off 2027: 15 Weeks Left to Find the Analogue Line

21 Oct 2026
same beat · Service OperationsOne Queue? The ITSM and Customer Service Convergence Nobody Can Ignore17 Oct 2026same beat · Service OperationsAI Agent Observability Just Changed Again, What Next?14 Oct 2026