If you’ve spent any time in Zendesk Admin Center lately, you’ve probably noticed the Zendesk App Builder feature.
It sounds pretty compelling: an AI-powered tool that makes it easy for anyone to create custom Zendesk apps, built specifically for your support team.
Better yet, it’s supposed to work for users without any development experience, because everything is no-code and uses natural language prompting.
But real life is often more complicated and nuanced than marketing hype makes it seem.
While some teams are absolutely finding real value building simple productivity tools and apps within Zendesk, that’s not always the case.
If you’re a support leader trying to figure out whether you should invest time in Zendesk’s App Builder or stick with third-party apps on the Zendesk marketplace, it’s important to get clarity on what’s working (and what’s not).
Let’s look at what teams are actually building with Zendesk App Builder, where projects succeed,
What Is the Zendesk App Builder?
Zendesk App Builder is an AI-powered tool that generates custom Zendesk sidebar apps through natural language prompts.
Put simply, it’s a tool that aims to enable anyone to create custom Zendesk apps for their team. For instance, you might prompt App Builder with, “Show me the customer’s VIP status and their last five tickets with color-coded priorities.”
App Builder would then generate the code in real-time, give you a live preview, and let you iterate by replying back to the conversation (just like in a Claude or ChatGPT conversation).
Once you’re happy with it, your new custom Zendesk app is ready to release into your Zendesk instance.
App Builder’s release was announced in October 2025, and it’s currently available on the Suite Professional and Suite Enterprise plans.
Main Features of the Zendesk App Builder
Building with AI through simple conversation
The core workflow of Zendesk’s App Builder is straightforward, and familiar to anyone who has used ChatGPT and other foundational AI models.
To get started, you just open App Builder in Zendesk’s Admin Center, select “Start a Conversation”, then select where you want the app to appear (sidebar, top navigation, or side navigation), and start describing what you need.
According to testing by Zendesk blog Internal Note, simple prompts like “I want an app that shows customer name, organization, email address, user type, and organization type” can generate impressive initial designs.
The AI processes your requirements, proposes a plan with core and optional features, and you can refine before it starts building.
Each response takes 30-45 seconds depending on complexity, so patience is part of the process (that’s still way faster than custom coding your own app!).
Preview and test in real-time
As the AI generates your app, you see a live preview using mock data. This gives you immediate feedback on the layout and basic functionality.
The App Builder platform includes version control, which means every change creates a new version with timestamps. That’s great because as you iterate, you always have the option of reverting back to previous iterations.
Before publishing your custom Zendesk app, you can test it in your actual Zendesk environment with real tickets and customer data. This is where the gap between “looks good” and “works correctly” often becomes apparent.
Connect third-party apps and internal systems
One of App Builder’s strongest features is its ability to let you integrate data from other apps in your support tech stack.
Your custom apps can connect to external services like Shopify, Jira, and Slack, using either API key authentication or OAuth 2.0.
AWS has documented some real success stories like Comcast, which built an employee profile merge app that saved agents over 5 minutes per merge by consolidating a 10-step process into two clicks.
That’s a perfect example of the power of App Builder to solve simple workflow issues.
The key word here is “simple” integrations. Apps that fetch straightforward data from a single endpoint tend to work well.
Complex multi-step API workflows? That’s where things get trickier.
Real Success Stories: When App Builder Shines
While there aren’t too many publicly available stories of App Builder’s impact yet, AWS’s other case studies highlight how App Builder excels at enabling support teams to create ic productivity-related apps that streamline unique workflows.
Comcast is one example, but others include:
- Catapult used it to automate RMA processing for 7,000+ annual returns, reducing ticket handling time by 2-7 minutes per request. That’s somewhere between 200 and 800 hours of time saved.
- Trust Payments built workflow automation that pulled transaction details straight into Zendesk. It unlocked a 50% improvement in efficiency, saving about 50 hours per month across their customer service team. Their Zendesk Workflow Manager calls App Builder “one of the best innovations I’ve seen from Zendesk.”
The pattern in successful projects is clear: they’re solving unique, internal workflow problems that no marketplace app addresses.
Great use cases include things like:
- A custom time tracker that logs hours to your specific billing system.
- Data viewer that pulls information from your proprietary CRM.
- A simple automation that fits your exact ticket routing logic.
These aren’t complex applications—they’re targeted tools that do one thing really well. More importantly, they’re tailored to your team’s workflow and tech stack.
When your need is this specific and relatively simple, Zendesk’s App Builder can deliver genuine value in hours instead of the months traditional development could require.
Why App Builder Projects Fail: the Reality Check
App Builder’s potential is incredible. Let’s be clear about that.
But for every success story, there are also frustrated teams who’ve abandoned partially-built apps after exhausting their prompt allowance. Understanding common failure patterns helps you avoid these pitfalls.
Complex data relationships break down
Internal Note’s detailed testing of App Builder revealed a critical limitation: App Builder struggles with complex Zendesk data structures.
During testing, they couldn’t get it to handle Custom Object lookup fields, which require chaining four different API calls in sequence:
- Getting the field value (returns an ID).
- Checking ticket_fields API to find which custom object it relates to.
- Querying the custom objects API for that specific record.
- Using custom_object_custom_fields API to get field labels.
That might sound complex, but the reality is that this isn’t an edge case. Many support operations teams use custom objects, ticket relationships, and complex field mappings within Zendesk. If your app requires things like this, App Builder can struggle to connect the dots.
The Gap Between Preview and Production
Another common challenge with Zendesk’s App Builder (also experienced in Internal Note’s testing) is that the design doesn’t always function correctly once pushed into a production environment.
App Builder might give you a really clean design with a logical structure and every element you need, but when you start testing with real data, functionality goes off the rails.
While you can work with App Builder’s AI to try and overcome these issues, if you’ve ever tried troubleshooting something with AI, you know it often involves a ton of back and forth.
Sometimes fixing one thing will cause another thing to break. It’s easy to spend hours going back and forth with increasing levels of frustration.
While many of these issues are fixable, that building frustration and cycle of breaking and iterating is a place where it’s really common to give up on App Builder.
Prompt Budget Becomes a Real Constraint
Your Zendesk plan dictates how many app conversations you get with App Builder (currently 5 on Professional and 10 on Enterprise).
Those conversations have a limit of 200 messages per conversation, and include up to 3600 prompts (with the option to request a one-time increase up to 7200 prompts, which may or may not get approved).
“Prompt” is also hard to define, because Zendesk’s documentation says that it’s dependent on length and complexity:
“The maximum number of prompts allowed is based on the expected length and complexity of each prompt. If you submit larger prompts, the total number of prompts you can send will decrease accordingly.”
That means going back and forth to repeatedly troubleshoot an issue with an app can easily burn through a lot of your prompts for the year.
Bottom line, you need to really carefully manage your App Builder prompt usage or you might run out of room to experiment.
Disadvantages of Zendesk App Builder
Beyond project-specific issues, App Builder has structural limitations worth understanding before committing time to building your own app:
- Not actually “no-code” for complex apps: While marketed as no-code, complex apps require understanding Zendesk API structures, data relationships, and authentication flows. You may need to review and edit the generated React code directly, which kind of defeats the “anyone can build apps” premise. That’s not unique to Zendesk or App Builder though; most “no-code” applications only get so far before they require more technical knowledge.
- Platform constraints: Apps can’t be moved between sandbox and production—you must manually recreate them. Once you delete a conversation, you lose the ability to update that published app. That limitation could cause a lot of regret.
- Maintenance is on you: Apps created through App Builder are “non-Zendesk services.” Zendesk doesn’t guarantee their functionality, support, or continued maintenance. If APIs change or Zendesk updates break your app, fixing it is entirely on you. That may require rebuilding from scratch.
For features requiring reliable, ongoing maintenance and things your customers and support reps rely on—like comprehensive help center analytics or automated ticket translation—professionally maintained marketplace apps provide better long-term value.
When to Choose Zendesk Marketplace Apps Instead
If you’re currently evaluating whether you should build or buy your next Zendesk app, here’s a helpful framework to think about:
- If it’s a simple, straightforward app that solves one problem, it’s worth exploring building.
- If your Zendesk instance is relatively simple (not many custom fields, integrations, etc.), building an app with App Builder might make sense.
However, if any of the below are true of your Zendesk instance and support team, you’re better off evaluating existing, professionally built and maintained Zendesk apps.
With apps on the Zendesk marketplace, you get the support of experienced app developers, you can see customer reviews and ratings to understand if an app actually works, and you avoid wasting tons of time and resources on apps that go nowhere.
Default to looking at existing Zendesk apps if you have:
- Tools that help you scale quickly: When you’re growing fast, you can’t risk custom apps breaking down when you need them most. For instance, take the Zendesk Agent Co-Writer. It does everything from summarizing tickets, enabling two-way multilingual support, and instantly generating on-brand draft replies to customer tickets. Taken together, these things make every support agent 10x more efficient—do you really want to risk losing all of that during your busiest time of year?
- Heavy data integrations: If your app needs complex, multi-step data fetching or real-time syncing with external systems, marketplace apps are better at handling that volume and edge cases you can’t anticipate For example, if you need robust help center management with bulk operations across hundreds of articles, Help Center Manager provides proven functionality that would take dozens of prompts to replicate imperfectly.
- Mission-critical workflows: Never experiment with App Builder for workflows your team depends on daily. If ticket automation with AI is core to your operations, something like Zendesk Triggers+ChatGPT offers production-ready no-code automation with guaranteed reliability and ongoing updates.
- Complex analytics and reporting: In its current form, App Builder struggles with sophisticated data visualization. If you need comprehensive metrics like article views, helpfulness scores, and self-service rates, Help Center Analytics provides insights that would take weeks to build and won’t break when Zendesk updates their API.
Experienced Zendesk app developers spend years building robust apps to solve common problems faced by most Zendesk customers.
It takes time and effort, because seemingly simple features have complex edge cases. App Builder can’t anticipate these scenarios.
Getting Started: Setting Realistic Expectations for App Builder
If you decide to try Zendesk App Builder, here’s how to set your project up for success:
- Start extremely simple: Build a single-feature app that displays data from one source. Test that it works reliably before adding more complexity.
- Focus on unique internal needs: App Builder makes sense when no marketplace app addresses your specific workflow (and the task is straightforward).
- Budget time for iteration: Even simple apps rarely work perfectly on the first try. Plan for multiple rounds of iterating to get your app to a workable state.
- Know when to stop: If you’re on your 15th prompt trying to fix the same issue, it’s time to either find a marketplace alternative or hire a developer.
- Document everything: Since you can’t edit apps after deleting conversations, save your prompts and any custom modifications somewhere safe.
The Bottom Line for Support Leaders
Zendesk App Builder represents genuine innovation in making custom development more accessible.
For teams with unique, simple workflow needs that no marketplace app addresses, it can deliver value quickly. The Comcast and Catapult examples prove that targeted productivity apps work.
And it’s far better than trying to build a custom Zendesk app with a tool like Claude Code or ChatGPT—neither of which really understand Zendesk’s data structure and intricacies in detail.
But the promise of “anyone can build anything” hasn’t materialized (yet).
Complex integrations, data relationships, and mission-critical features are still a better fit for proven apps from the Zendesk marketplace or for working with a Zendesk developer.
As a support leader, your time and your team’s efficiency are valuable. Before investing weeks into building with App Builder, ask:
- Does a proven marketplace app already solve this problem?
- Is our need simple enough that iteration won’t consume 50+ prompts?
- Do we have the technical knowledge to troubleshoot when things break?
- Are we willing to commit to maintaining and updating this app long-term?
If you’re uncertain, or if you’d rather start by exploring proven solutions for your support team’s specific challenges, you should book a demo to see how Swifteq’s suite of Zendesk apps can streamline your customer support without all the painful trial and error.
Written by Larry Barker
Larry has spent over a decade leading CX teams at tech companies of various sizes. He also currently operates Supported Content, a niche content marketing company that helps CX brands attract and retain customers.
