Welcome to the final Internal Note of the year. 52 weeks, 70 articles and over 100.000 words later we’ve ended up at this year in review article, wrapping up an amazing year for me at Internal Note.

The highlight professionally for me this year has to be Relate last April. An amazing week of meeting and talking to Zendesk Product members together with my namesake at Premium Plus, Thomas D'Hoe.

That and managing to double the audience of this blog, and seeing that number rise each month. So to all readers and subscribers, both free and paying: thanks for reading! Couldn't write this website without your feedback!

Last year I wrote a more personal review of the year, but this year I did something different and wrote a real overview of everything Zendesk announced over the year.

To make sure my overview was not too biased, I reached out to a dozen friends in the Zendesk ecosystem to give their input on Zendesk’s year, which result in a nice scorecard for the Zendesk Platform. So throughout this review you’ll see a score for each of these pillars, which is based on not only my opinion, but also those of those people I reached out to. And for those who filled in the survey: thanks!

SweetHawk apps for Zendesk

Apps that power Employee Experience in Zendesk® & much more!

SweetHawk apps for Zendesk0

Major themes of 2024

Since yearly reviews can’t contain every little thing Zendesk announced, I’ve decided to focus on three major pillars:

1. Customers and Employees: Messaging, AI Agents, Help Center
2. Agents: Agent Workspace, Advanced AI
3. Platform: Data and Security, Marketplace, Developer APIs

Before we dive into the details, let’s set the big picture first. When we look back at 2024 I think we can see 5 big themes that guided Zendesk innovations this year.

The 5 biggest themes from Zendesk this year are:

• 🤖 AI Expansion: Zendesk is heavily investing in AI, evident in the acquisition of Ultimate.ai and the launch of AI-powered features like AI Agents, Agent Copilot, and generative AI for various functionalities, including voice, summaries and intelligent triage.. These advancements aim to automate customer interactions, enhance agent efficiency, and provide data-driven insights.
• 💰 Shift in pricing: Moving away from traditional seat-based pricing, Zendesk is introducing outcome-based pricing for AI Agents, where customers pay for resolved conversations. This combined with the new Dynamic pricing that allows customers to shift agent seats towards Automated Resolutions allows customers to adapt their spending and invest in the platform features that help them give the best results. AI Agents with its built-in ticket deflection and context gathering, or human agents supported by Agent Workspace and Advanced AI to give more personal support to more complex issues.
• 🔗 Unified Experience: Zendesk is working toward a unified experience platform that seamlessly integrates AI, automation, and human agents. Features like Omnichannel Routing, Messaging Authentication, and the new Zendesk Voice aim to streamline customer interactions across all channels, offering a consistent and efficient experience.
• 🧑🏻‍💻 Focus on Employee Experience: Alongside customer experience, Zendesk is emphasizing employee experience, recognizing the impact of agent well-being on service quality. The acquisition of Klaus and the development of Zendesk WEM, which includes workforce management and quality assurance tools, demonstrate this commitment to supporting and empowering agents.
• 📊 Data-Driven Insights: Zendesk is enabling data-driven decision-making by providing deeper insights into customer interactions and agent performance. Enhancements to Zendesk Explore , the introduction of Intelligent Triage with Entity Detection, and advancements in Zendesk QA equip businesses with the data they need to optimize their support operations, identify trends, and improve customer satisfaction.

In 2024 also saw the launch of the Customer is always human slogan, which evolved over the year to AI-first service, catered to humans. I really like this reworded second version cause it focuses not only on customers, but also implies employees, and agents.

Year in review

Looking back at 2024 I was astounded by the amount of new stuff Zendesk released. I started noting down the biggest items but quickly ended up with a giant list of features and no real structure for this article. Similar to how I resorted last year to Slite to get an insight in my articles and Zendesk’s releases, this year I decided to take a look at Google NotebookLM to get some insights and summaries of the last 12 months.

So for those who prefer to listen to a Zendesk year in review while preparing their turkey or other Christmas meals instead of reading an article with a nice cup of coffee, here’s a podcast generated by Google NotebookLM about Zendesk in 2024. I have to admit, after importing all articles and generating the podcast I was astounded by the quality of the output. 

But no worries, I won’t move to using AI for this blog anytime soon. AI’s good for insights, but original work is still something we humans are better at.

Zendesk Releases in 2024 - An AI Generated Podcast

0:00

/1604.6

1×

Customers and Employees

Messaging, AI Agents and the Help Center

Messaging

Will 2025 be the year Chat finally dies? Looking back at 2024 it’s clear that Zendesk Messaging is at feature parity with the old Zendesk chat. This year we got a migration to a new messaging backend that integrates the old chat triggers into the admin center. These same triggers got expanded with new options like an Average Wait Time trigger as well the ability for agents to end chat sessions and give control back to the customer or AI Agent.

Speaking of giving control, the new Multiple Conversations ability in the web widget is a long requested feature. It allows customers to have multiple conversations concurrently but it also fixes a couple of other issues.
One of those is the “restart conversations” capability that’s been often requested. When you’re stuck in a flow or want to reset the conversation there didn’t really used to be an option for that. The old Ultimate widget had that feature, and the legacy Zendesk widget allows you to close the browser tab and reset your session, but Zendesk’s continuous conversations feature prevented you from doing that.
With the new conversations button it’s easy to jump out of your current chat and start a new one. And since you’ve got multiple conversations visible in the widget it’s also easy to switch from talking to the support team to gathering more context about an unrelated topic from the bot.

A big part of this is enabled by the expanded support for authentication for end-users. The new verified users option makes sure you’re talking to the right customer, and also allows you to use that user metadata in bot flows, greeting messages, or showing older conversations to logged in users.
The logins are still based on an external JWT endpoint though, and I really hope 2025 will offer a way to easily inherit signed in sessions of the Zendesk Help Center, or offer customers a way to authenticate during an active conversation, instead of only being able to login before you start talking in the Zendesk widget.

Messaging

Plenty of quality of life improvements that make Messaging a lot easier to implement and more feature complete when compared to the old Chat.

One element of Messaging I hope to see improved in 2025 is the Zendesk widget. Even though Zendesk's bots are rich and powerful, the widget itself is beginning to feel a bit stale. Compared to Intercom's widget for example Zendesk's widget feels a bit empty. Where other players in the CX market link to blog articles and help center articles in a addition to chat and voice, Zendesk's widget is purely focused on chatbot interactions.

Lots of updates on Messaging, almost feature parity with Chat.
(Ana Luisa Santos - Premium Plus)

AI Agents

Zendesk Messaging and Chatbots go hand in hand. And AI Agents – as Zendesk calls their bots now – are one of the parts of Zendesk that got the most attention of all features this year.

The legacy Zendesk Bot, which is shortly getting replaced with an Ultimate powered version, got a lot a lot of quality of live improvements this year. Flow builder got support for linked answers, parameters, more powerful conditional flows and dynamically generated carousels and option lists. 

The legacy Zendesk Bot itself also got support for generative replies and persona, turning your Help Center articles into short context aware responses to customer inquiries. These Generative replies based on your content offer more personal direct solutions to customers, while making sure you can leverage your articles across the Help Center, email, and conversations.

The big announcement this year however has to be the acquisition of Ultimate. Not only did Zendesk get a much more capable conversational AI model in house, it also opened the door for Zero Training custom use-cases, more powerful API integrations, hybrid flows that combine generative AI with traditional step-based answers and support for a lot more languages across the board. 

However, with the acquisition Zendesk also created a lot of confusion. Zendesk Bot got renamed into AI Agents, and AI Agents became an umbrella term for the legacy Zendesk Bot and the more advanced Ultimate AI Agent. While both bot platforms are kept alive with new features in parallel this year, I hope 2025 will bring a unification of both platforms. 
If you’d ask me I expect the Zendesk Bot to disappear, begin replaced with a Ultimate-powered solution for both customers that only have Zendesk Suite (with basic features), with a more powerful add-on for those that need more. Time will tell, and with Relate 2025 quickly approaching in March, it’s only time before we see what’s going to happen here.

AI Agents aren’t just about chatbots though. At the AI Summit we heard that 60% of customers still contact support teams over classic channels like email and voice. 
Zendesk’s autoreply (or answer bot as it used to be called) got renamed this year but basically got zero improvements, while Ultimate’s email automation got support for generated replies over email and an early release for conversational support across multiple replies, where the bot can interact in multiple back-and-forth within the same conversation. 
Now that Ultimate’s bot can detect entities like an order number in email, and generate responses based on customer questions or those entities, using generative AI as a solution to deflect and self-service over email channels became a lot more powerful.

Aside from chat and email we also got a sneak preview of the improvements for Voice Agents at the AI Summit. I’ve talked to the team at PolyAI and seen their product in action, but since the AI Summit we haven’t really seen any movement on the Voice Bot front, so I hope we see more in 2025.

Wrapping up the AI Agent section, we’ve got maybe the most controversial change of the year with the introduction of Resolution based pricing. Zendesk traditionally has always been a licenses based platform. You’ve got 50 agents, you pay for 50 seats. (Light agents have always been a wonderful exception to this rule). With the acquisition of first SunCo, then Ultimate and the launch of Answer Bot in all its variants, they’ve played around with concepts of Answer Bot responses, Monthly Active users and other variants that didn’t really went anywhere. This year they did seem to get all their ducks in a row and launches the new Automated Resolutions for AI Agents. In essence it comes down to: if the AI Agent does its job and resolved a ticket, then you pay for that action. If the ticket goes to an agent, you don’t.

It’s a big shift away from the agent based pricing we’re used to, and putting the fairness of such pricing aside, does impact your budgeting a lot. Where previously you only needed to account for people, you now need to predict or take ticket workload and your AI Agents’ efficiency into account when budgeting out your CX platform in order to decide on the right mix of agent seats and automated resolutions.

Luckily the launch of dynamic pricing takes that into account and offers customers the ability to dynamically change the balance of seats and bot actions in your contract based on current needs. This is especially useful if you’re on your way to a higher automation rate, moving from pure generate responses towards more complex hybrid bot flows and API integrations. Where the former will deflect 10-20% of your tickets, leaving a bigger workload for your agents, the latter can go up to 40-60% of automated tickets, lowering the amount of tickets for agents (and thus reducing the required headcount) while increasing your AI Agent cost. It’s a tricky thing to balance, and I’m not in a position to judge if Zendesk’s pricing is fair, realistic or future proof. But with OpenAI launching a 20 or 200$ a month tier for ChatGPT, Microsoft asking up to 50$ a seat for Copilot, and others charging too for their AI work, it’s clear that paying for AI usage is only natural.

AI Agents

If you know me you know I’m a big fan of Ultimate’s product. It’s rough around the edges and could use some Zendesk Garden UI attention to detail, but the platform itself is o so powerful. I do hope Zendesk will cleanup their naming and offering next year but I do expect some confusion and frustration once the new AR based pricing becomes common for all renewing customers.

The Ultimate AI Agents are very supportive with ticket automation, the integration works very good and starting up your support bot with a help center up to date is very easy, the generated answers are providing links to the source on your help center.
(Rob van den Boogaard)

Help Center

A successful customer service approach is defined by two key elements: having good documentation in place that allows customers to self service and resolve issues themselves, and having the right processes in places for when an agent does need to jump in.

The Zendesk Help Center is the tool to assist you with the former of these items. And where it used to be the sole way customers could find answers to their questions, Zendesk Guide has evolved more and more into a good place to manage your knowledge, with the Help Center being just one of the ways customers and agents interact with your content. Where searching an FAQ and reading an article was the primary interaction model up to a year or two ago, more and more we’re now evolving into a world where articles are the basis for custom generated and personalized responses to your customers in AI Agents over chat or email.

The same is now also true for the search function on your Help Center. The new Quick Answer feature shows an inline generated response to a customers’ search query above the actual list of articles, giving them immediate answers to their questions.

All these new ways of exposing your content to customers also mean you’re pressured more and more to keep that content up to date and relevant. A big focus was put on the editor experience to make that work easier to do. The article editor got a new sidebar that streamlines settings, a less cluttered editor window and the release of generative AI tools like simplify or tone shift.

Where the text of articles is more and more applied as the basis for generative replies, the media content of your documentation remains a unique benefit of the Help Center. Embedded photos, videos and attachments can make a boring support article that more clear (and fun) to read. One of the underlying features in the article editor, the image library got a massive overhaul this year with a centralized gallery across all articles, the ability to update images across articles, upload locale specific versions and bulk import assets from external sources.

Wrapping up 2024 was the release of multiple segments for an article, finally allowing you to publish one article in multiple sections (with multiple brands being rumored as a release for next year)

Help Center

My biggest gripe with the Zendesk Help Center is its content editor. The editor got major improvements this year but it still acts like Word. I’d love to be able to write markdown and have it convert automatically to titles, lists or links. Embedding images and tables with a / command is easier than clicking menu options, and embedding links, social cards and video embeds would be quite nice. I might be too used to the editor in Ghost, the platform I use for this blog, but writing in Zendesk Guide always feels a bit less smooth.

Aside from that gripe I do like the update of the editor design, it’s less cluttered than before and options are not the main thing to focus on when looking at that page. 

Solid. But SEO and customizability out of the box could be improved.
Nils Rebehn - Guidoo Services 

Agents

Agent Workspace, Agent Copilot and Advanced AI

The last year(s) at Zendesk are years of relentless innovation and feature releases across the entire suite of products. With the above overview of customer facing releasing I’ve barely done justice to all the features and improvements that have been released, touching on both big topics like AI, but also small qualities of life like ending sessions.

A similar story is true for the Agent side of Zendesk Suite. To structure this overview, I’ll split this section up in two parts. The first focusses on the classic Zendesk experience for agents in Agent Workspace. The second will round up all the AI innovations integrated into the Agent Workspace.

Let’s dive in.

Agent Workspace

Zendesk has traditionally been a very static trigger-based setup. A ticket comes in, triggers update the ticket, and the ticket ends up with a specific group of agents. But over the year Zendesk has gradually (and then very suddenly) started moving everyone over towards a more dynamic routing experience where tickets are routed to the right agent.

Omnichannel routing takes elements like locale, topic, channel and customer into account to route tickets to an agent who has time, availability and the right skills. Queues can add an additional layer of triaging to make sure the right tickets are handled first, and Agent Home is the new default location where agents can find their work.
Not every customer runs on Omnichannel Routing already, but now that it’s the new default for any new customer, those that still use their old trigger based approach will quickly feel that the amount of improvements on “the other side” will make it the more logical choice sooner rather than later.

I glanced over it in the previous paragraph, but the Agent Home is the one element I see the most potential in for Agent Workspace. Currently it’s a nice overview of your assigned work, but the more (AI powered) insights Zendesk adds on top of tickets, the more I see the Agent Home evolve as not only a way for agents to see what’s to do, but also a way to find out what’s up.

Once we dive into the tickets themselves there’s also a lot of changes. The conversation view that used to be very email-like has now moved to a conversational view with smaller text balloons and less repeated header text. Macro suggestions, Side Conversations, user context cards, custom objects cards e.a. have all been moved into the Context panel, giving you a more focused view on the conversation itself. And within that conversation you now have the ability to end conversations over messaging (pushing the customer back to the AI Agent), and a richer view of conversations with that AI Agent with nicely rendered buttons and carousels.

In between Agent Home and the Ticket sit Views. Views give a clean list of a specific set of tickets based on criteria you choose. And even though personally I’m not a fan of long list of views, the new view categories with its collapsable sections at least make the overview tolerable. The ability to use Followers and CCs as conditions in Views (as well as their appearance in the Agent Home made a Marketplace app of mine, Follower View, useless, but it’s awesome to finally see these conditionals appear as a native options in Views.

Agent Workspace

Scoring this one was tricky, since I’m so used to the AI powered version of Agent Workspace as described below. But purely looking at the classic experience, the biggest impact for me is Agent Home and the new Conversation View.
Agent Home does require a complete rework of your Zendesk to get the most out of it, and the new Conversation View, although a major change from the old look, works so much better when talking over Messaging and to see Bot interaction history that I wouldn’t want to go back.
I do think it’s great that even 15 years in, Zendesk still dares to rework major parts of their platform and the way it works. I’d love to be able to use Omnichannel Routing without that stupid tag, and queues miss some major insights (like a view of queued tickets), but it’s clear where the platform is going and like I said, I love the fact that they dare changing such fundamental things.

Great overview for agents, with all new functions we can integrate more into the workspace so the agents have all in 1 screen to support the customers.
(Rob van den Boogaard)

Advanced AI

When it comes to naming their AI solutions, Zendesk’s naming is equivalent to Sony camera names. Illogical, inconsistent and chaotic. Macro Suggestions and suggested macros. AI Agents, not to be confusion with AI Agent Copilot. A Copilot that used to be feature, and is now an overall name for a bunch of features. And that’s before we talk about AI Agents that exist in two flavors, and somehow are still called Zendesk Bot too. 😅

Either way, where 2023 was the year Advanced AI was released, I think we can agree that 2024 was the year where it became an add-on worth its price.
Not only did we see the releases of tone shift, summary, suggested macros, similar tickets, merge suggestions and suggested replies, we also got both the announcement and release of the Agent Copilot.

For me Copilot will probably have the biggest impact on agents’ experience moving forward. The combination of procedures, actions and suggested replies pulled from the Help Center almost immediately upon launch shows its promise. Procedures allow Zendesk to suggest replies to your agents to gather feedback from the customer (like missing order number), or prompt replies based on the status of the ticket (we’re working on it!). And actions take this one step further by offering agents a one-click button that executes actions in external systems.

Last year Zendesk launched Advanced AI with the addition of an Intelligence Panel to the side of tickets that showed intents, sentiment, summaries and all of the other AI powered features in Zendesk. One piece of feedback I heard from a lot of customers of mine is that agents now had to choose. Either show the added context provided by AI, or show their apps and customer context. The result was often that agents ignored the Intelligence Panel and went back to the tools they already knew.

With the launch of Copilot at the AI Summit, Zendesk also launched a revamped Agent Workspace that integrated the elements of Advanced AI more deeply in ticket views by adding summaries and intents to the title section, integrated suggested macros in the macro dropdown, and moving similar tickets to search. This basically turns many existing features in the Agent Workspace in an AI powered version of those tools, turning a familiar environment for agents into a more powerful version, without breaking any muscle memory by keeping the tools in the same place.

To wrap up this years’ releases for Advanced AI we saw the addition of transcripts and summaries for voice, and an extended version if Intelligent Triage that now not only offers a generic Zendesk AI model for all AI customers, and the ability to define custom Entities that detect items like products, locations or order numbers in tickets.

Advanced AI

In the previous section I spoke about a classic experience for Agents. I think moving forward we’ll see a dichotomy in Zendesk of environments that have a classic experience with triggers, macros and ticket fields, and an AI powered version of Zendesk with intelligent triage and auto-assist. The former is a proven solution for more than a decade, the latter a more powerful version of the Zendesk we know. 

That more powerful version comes with a few caveats though. Copilot can one be successful if you’ve got a solid set of procedures and a way of working that allows agents (or Zendesk) to make decisions, and have a tech-stack that allows for integrating and automating actions like doing a refund or rebooking a flight. If your setup is fractured, your agents limited in the decisions they can make, or your external tools and departments isolated from your CX team, Copilot won’t be successful.

I did give AI for Agents a lower score than AI for customers. The main reason is the limited scope of Zendesk intents’ model. If a company’s use cases don’t fit within Zendesk’s model, your intent tagging will be a bit too generic. The painful way in how we can link intents to queues and triggers also needs a rethink. No one wants to go through a dropdown of hundreds of options and select all valid intents one by one. 

The lack of Zendesk native actions for Copilot is also limits Copilot a bit. Updating ticket fields, adding tags, setting custom statuses or creating side conversations would all be first-party actions you’d expect to be there no?

Copilot is promising - but we'd like to see more actions.
(Geoffrey Brocard - Premium Plus)

Platform

Data and Security, Marketplace and Developer APIs

Zendesk is a platform that has customer, agent and company facing elements. And even though most of the marketing and customer attention goes towards the bots and AI features , a big part of Zendesk is hidden within the Admin Center and APIs. 

What I really like about managing Zendesk instances is that most of what can be done via de Admin Center and Agent Workspace, can just as well be accomplished via API. Nearly all interactions can be automated via webhooks, apps built for Zendesk can interact with tickets, and all your data can easily imported and exported.

Data and Security 

With the arrival of GDPR and similar focus on privacy giving companies controls to manage data efficiently is key.
Zendesk’s platform got plenty of new features that allow you to do just that, with Zendesk taking a carrot and stick approach.

The first carrot focuses around data retention with the addition of rule sets that will delete a subset of tickets after a set amount of time. Customers who bought the Advanced Data and Privacy (ADPP) add-on get more customizability and additional rule sets, but it’s great that everyone gets a “set once and forget” auto-delete option to clear out old tickets.

Similarly, similar to how Advanced AI’s entity detection reads and highlights specific elements in replies, the new automated detection of PII and bulk redaction features in ADPP make removing unwanted (and potentially dangerous) personal information from tickets an breeze.

Tickets aren’t the only data type in Zendesk though. Users, organizations and custom objects take up an equal amount of space in your instance. Last year saw the arrival of the new Data Importer for Custom Objects, and this year this new tool got expanded with support for importing Organizations, and got support for create, update or create and update capabilities. I assume it’s only a matter of time before we can manage Users via this new importer too.

While importing data became easier with the new importer, deleting the same data changed a bit too. Over the last year Zendesk’s been gradually moving end-user management out of the Admin Center into the new Customer tab in Agent Workspace. (And similarly has been moving agent management into the new Team tab in Admin Center). Elements like suspended users, (bulk) deleting customers e.a. are now all nicely centralized into this new location.

Zendesk’s new data management features come with a big caveat though. Starting this year Zendesk is imposing Storage Limits on all accounts, giving an overview of your usage in the Admin Center (together with API limits, Automated Resolutions and additional macro and trigger usage reports), and will start charging you when you go over your allotted amount of storage. By leveraging the new retention rules you can control your data usage, while also removing old data making GDPR compliance and basic privacy etiquette simple.

Data and Security

Personally, I'd give this a 4 out of 5 if more of the bulk deletion and redaction features were available to all customers instead of being part of ADPP. Security and Privacy aren’t gated by the size (or budget) of a company but should be available to all. 

However, I do appreciate the continued work on making data management in Zendesk easier, and seeing some really old features like organization import and user deletion being gutted and replaced with modern alternatives is nice to see. It proves the company is keeping an eye on not only the visible features like AI and Agent Workspace, but also dares to touch and change the fundaments of the platform.

Marketplace

Changes to the Zendesk Marketplace by Zendesk in 2024: it got a redesign.
That’s it.

All kidding aside, there's nothing new on the Marketplace from Zendesk's standpoint aside from maybe the Relay app. When looking at what partners delivered from a themes and apps standpoint, there's a lot of great new stuff out there.

Marketplace

Zendesk has a vibrant ecosystem of technology partners that build on Zendesk to integrate, expand and enable workflows and processes for its users. Marketplace themes allow you to make your Help Center themes fit your brand, and the Partner overview allow you to find the right implementation partner to assist with implementation.

If you look at the sponsors if this blog alone, you’ll find tons of quality solutions available:  tools for employee services via Sweethawk, complex form and fields management via Cloudset, advanced views and filters via Lovestock & Leaf, workflows across teams and tools via Next Matter, AI powered automation via Knots, an advanced voice channel via Babelforce and powerful tools to manage your Help Center via swiftEQ to name just a few.

It’s a shame that the marketplace, which is one of the main ways partners can show of their apps to customers feels so old. This quote from one of the Marketplace partners sums it up nice:

Not much has changed. There is still a bunch of stuff they should improve like being able to play videos, sponsoring adds for apps, getting more data on what people are searching for etc.
(Partner on the Marketplace)

Developer APIs

The developer documentation of Zendesk has to be one of my most used webpages while working for Premium Plus or writing for this blog. It contains rich documentation, good examples and somehow, even after all these years, still has API calls or hidden parameters in specific endpoints that make importing data for a customers or implementing an integration that much easier all of the sudden.

Similarly to how Zendesk making big changes to data importing and deletion is a good sign, for me a healthy API that stays in sync with features additions is an equally good indicator of the health of a platform. By releasing features in the admin center and making APIs available to interact with those features at the same time it shows that those features aren’t just “hacked” into the front-end but are structurally integrated with the entire platform.

This year saw the releases of APIs for Queues, Redirect rules for the Help Center, a new Help Center template with support for end-user lookup fields and AI powered Quick Answers, New APIs to manage for agent workload and messaging conversations, bulk importer APIs for Custom Objects, Media management in the Help Center, an API for the new Customizable CSAT survey and many more tiny updates to existing APIs. 

Developer APIs

Zendesk’s API have to be one of the nicest I’ve worked with. There’s room for improvement though. I’d love to see a more granular approach to API tokens that only give permissions to a subset of data. The complexity of having users live in Zendesk and Sunshine Conversations makes managing user authentication and identification for messaging tricky. I’d love an API to import, export or modify AI Agents (the Zendesk kind) and I hope there’ll be an API to interact with Zendesk AI somewhere in 2025. Generating a summary via API to use in integrations, or doing entity detection on side conversations or within a sidebar app would be pretty nice to have, to name just a few.

Props to the dev teams that increased the clarity of the docs a lot. But also a big thanks for the collaboration on the custom objects API -
(Kay Heunen - Sparkly)

Wrap Up

With an average score of 4/5, I think we can conclude 2024 was a rather good year for Zendesk.

The highlight of the year for me was probably the new AI powered Agent Workspace due to the way the AI is integrated within the interface instead of drawing attention to itself with some cheesy sparkles and glow. AI is a tool to be used, not something to show off.

Close contender for me is Agent Home. It forces a complete rethink of the entire routing and assignment approach of a Zendesk instance, but the result is a lot cleaner and more efficient to work than a My Tickets view.

Looking forward at 2025 there’s a few items on my wishlist. I’d love to see an answer to the multiple AI Agents problem we have now. Zendesk’s AI Agent offering, both from a technology as well as a naming perspective is too confusion to explain with competing technologies that offer similar solutions.

I’d also love to see a more radical approach to deprecating old technology. There’s no gaps between Zendesk Chat and Messaging anymore, killing Chat would force customers to adopt a newer and better Zendesk, and makes live as a consultant a lot easier. And while we’re talking about killing old stuff, a more native way to enable Omnichannel routing and Agent Copilot that’s not just a tag would be nice too. The current way feels a bit like a hack to me.

And finally, I’d love to see a rethinking of Users in Zendesk. The combination of users and identities in the API, the fact that users have multiple accounts across social media that should be more easily linked to Zendesk, a way to differentiate between an Agent working in Zendesk and that same person being an Employee (and thus end-user) in a company, and a way to have user profiles linked to a specific brands are all items on my wishlist.

This, and a way to authenticate users within the Zendesk widget based on social profiles or Help Center authentication instead of the external JWT stuff we have now would also be pretty nice to get Zendesk…

So, there we have it. Hopefully this overview of Zendesk gave you some insights on what’s possible on the platform now and maybe inspired you to adopt some of the new releases in your instance in 2025.

If you disagree with any of the scoring, or feel I missed a major feature, do jump into the comments and let me know!

Happy holidays 🎄🎄🎄

Thomas.