Why B2B Teams Are Leaving Zendesk: Real Stories from AssemblyAI and Aptible

Many B2B support teams are still relying on legacy platforms like Zendesk. But these tools weren't designed for Slack support, native AI workflows, or the complex issues that technical support teams often get.

In our webinar, Ryan Seams (VP of Customer Solutions at AssemblyAI) and Ben Odom (Director of Customer Experience at Aptible) shared their experiences moving from Zendesk to Pylon. Both companies were longtime Zendesk customers before making the switch.

If you missed the live event, here’s a recap of the key insights, takeaways, and audience questions.

Why Zendesk Wasn't Working

Limited Integration with Slack

AssemblyAI’s five-person, global support team was juggling three separate channels: email, live chat, and Slack.

They started having challenges with Slack support in particular. "A lot of our biggest customers were pulling us there," Ryan explained. "They wanted a way to async communicate, feel like they're part of a team."

But managing hundreds of Slack channels manually wasn’t sustainable. Support engineers had to monitor up to thousands of individual threads, with no unified system to bring it all together.

Zendesk's Slack integration didn’t help: It was a one-way notification bot that couldn’t actually improve the team’s daily workflows.

Few Ways to Act on Account Data

When Aptible looked for alternatives to Zendesk, they were also focused on solving their Slack scaling problem.

But during a trial period, the team realized Pylon had many more features that would help them act on customer data. "It became clearer and clearer that we were going to be able to make a lot of use out of [our account] data that we weren't before," Ben explained.

With all their customer context in Pylon, they’ve tapped into new workflows they didn’t even expect: managing omnichannel support in one place, getting rich account insights while answering tickets, and dramatically improving their knowledge base management.

Evaluating a New Support Platform

Both leaders emphasized that choosing a support platform has fundamentally changed with AI. Here’s how to evaluate support platforms today.

Non-Negotiables

AI-native architecture, not bolt-ons. "Buyer beware of those features," Ben cautioned. "Make sure you've tested. Make sure they're not just bolt-ons." The AI tooling needs to be fundamentally baked into the platform, not added as an afterthought.

True omnichannel support. This means you can manage Slack, Microsoft Teams, email, live chat, and other customer-facing channels all from one unified system. The key word is "managed," not just "monitored."

Flexibility in workflows, SLAs, and data models. "If you have that and you have smart people, they can go build everything else on top of it," Ryan emphasized. Rigid systems limit what support teams can accomplish.

Integration depth. Ryan pointed out that many platforms claim to have integrations your team needs. But it’s important to make sure those integrations actually have the functionality to improve day-to-day work.

The Testing Process

For Ryan and Ben both, it’s important to do a hands-on evaluation:

• Put the tool in the hands of your top performers and change leaders
• Let them provide specific feedback on workflows and UI/UX
• Run a phased pilot, starting with lower-risk channels
• Measure the platform’s impact on daily efficiency before committing

"The best thing is to try it and experience it for yourselves," Ryan said. "The immediate feedback from our support engineering team was just great. 'My life is so much easier. I can actually go eat lunch in peace now.'"

Tips for the Migration

One of the scariest parts of switching platforms is the migration itself. Both teams approached it slightly differently.

Going Team by Team

Here’s how Aptible went about the migration:

1. Conducted a comprehensive gap analysis, documenting features by frequency and priority
2. Phased migration by team, starting with billing and security before bringing in higher-volume support channels
3. Imported two years of historical tickets to train AI in Pylon
4. Maintained Zendesk license temporarily, in case they needed to look at old tickets

"We did a test migration, found a couple issues of a small sort, and then were able to wipe and restart," Ben recalled. "But it was very straightforward. Migrating two years of tickets was way faster than I thought it would be."

The key was identifying power users early and getting their feedback to smooth issues out before the full cutover.

Going Channel by Channel

AssemblyAI phased their migration by channel:

1. Started with Slack, which had the lowest risk because they weren’t using existing Zendesk features there
2. Moved to live chat next, where they just had to change the code snippet on their website
3. Tackled email last, which was most complex because of routing rules

"You have to build momentum and build iteratively through these different phases," Ryan explained. "Show your end users: here's what we've done, here's the benefit to you, here's all the time you've saved."

For teams with technical resources, Ryan noted that moving knowledge bases and historical data can be done quickly: "You could literally vibe code something that would move your data from Zendesk to Pylon in less than a day."

Unexpected Benefits After the Migration

Both Ryan and Ben’s teams ended up getting value out of Pylon that they weren’t expecting.

A Customer Intelligence Hub

"We bought Pylon to be a support tool and facilitate customer conversations," Ryan said. "But what we ended up stumbling upon is that if you're able to send all of your customer data into Pylon and attach it to different accounts, you could actually start to understand trends."

AssemblyAI now pipes in:

• Usage data from Looker
• Error alerts from Datadog
• Product feedback from customer calls
• Support conversation history

This creates a unified account view that helps them identify at-risk accounts and upsell opportunities.

AI That Actually Works

Ryan's team deployed their first AI agent about a year ago. Today, it answers 35% of incoming tickets, "effectively something like three full-time employees," he noted.

The key is the feedback loop: when the agent can't answer a question, it escalates to a human. The support team then reviews gaps weekly and updates documentation, making the agent progressively smarter.

Ben's team took a different approach, keeping humans in the loop but using AI to increase their efficiency. "We touch all of our responses right now, but the assistance that the knowledge management and AI tooling gives us is amazing," he explained.

The team's CSAT scores have stayed high through the migration, while they've been able to:

• Manage Slack support for over 85% of their customers
• Generate and maintain knowledge base articles faster
• Get focused time back for platform improvements and bug fixes

Cross-Functional Impact

Thanks to their Pylon setup, folks on Aptible’s finance team can respond to billing issues directly from Slack. The customer success team also uses Pylon for account management, including using AI Web Research to pull customer news and provide proactive support.

At AssemblyAI, customer success managers live in Pylon and get alerts from multiple systems centralized in one place.

"The role of support becomes very much just working with customers and unblocking them, and reporting back all of those rich insights to the rest of your company," Ryan observed.

Key Takeaways

1. Legacy platforms weren't built for how B2B teams work today. You need tools that are designed for Slack support, AI workflows, and account-level intelligence.

2. Omnichannel support means unified management. Your support engineers shouldn't need to monitor hundreds of Slack channels manually.

3. AI features need to work natively. Test AI tooling to make sure it’s actually useful in your team’s workflows.

4. Phased migrations reduce risk. Whether you go team by team or channel by channel, building momentum with early wins makes change management smoother.

5. The right platform can become your customer intelligence hub. When all your customer data flows into one place, you unlock insights that would have taken hours to compile manually.

Top 4 Questions from the Audience

Q: Did you recreate your incident management workflows? Does it function better or the same?

Ben: "We've improved ours quite a bit… We've added some intelligence around when to escalate to PagerDuty, not just one static rule. We've added some AI-based escalations where if someone says something, a tag will be made and someone will get escalated to. So I would say we're using some features that simply were not available to us before."

Q: Can we configure automatic first replies that could actually help customers? This is one of our most important SLAs.

Ryan: "Yes. Effectively, that’s what our AI agent does. You can decide based on the category of ticket, the topic of the ticket, etc., exactly how you want that agent to reply. Imagine if you have it set up to handle billing tickets different from product tickets different from incident tickets, you can go and customize that.

So the short answer is absolutely yes. And you can do more than a first reply if you want to. You can have it handle the ticket end-to-end. It's really up to you how you want to configure that."

Q: Did your end users have any trouble adapting to the change in workflow? We have some CSMs who are stubborn about any change in process.

Ryan: "Of course, lots of people don't want to change… You have to find the people who do want change and have them kind of lead the way. Have them show the rest of the team: 'Here's how moving to Pylon from Zendesk made my day 10x easier, 10x more efficient, 10x simpler.'

There’s a lot of different ways you could approach this, but ultimately, people will change if they see that there's a benefit to them and a very clear value to that switch."

Q: Can our current Zendesk ticket history and knowledge base database be imported and referenced when working on tickets?

Robert (Pylon): "Yes, very doable. We have that migration option available with all the tooling, and we have a flow to migrate in the historical as well as the ongoing email tickets."

Leave Your Legacy Platform

Platforms like Zendesk weren’t built for Slack support, AI-native workflows, or unified account intelligence. If you’re thinking of switching to a new support system, phased implementation, hands-on testing, and strong support from your new vendor can make it a smooth transition for your team. You might even discover new features and capabilities for your support ops that were never possible before.

Pylon is the modern B2B support platform that offers true omnichannel support across Slack, Teams, email, chat, ticket forms, and more. Our AI Agents & Assistants automate busywork and reduce response times. Plus, with Account Intelligence that unifies scattered customer signals to calculate health scores and identify churn risk, we're built for customer success at scale.

Book a demo today.