Customer Support

Identifying which support issues to automate

After AI structures and analyzes ticket data, teams use the resulting visibility to identify recurring request types, knowledge gaps, and the highest-volume support issues worth automating first, with sample responses and savings estimates helping prioritize what to automate.

Why the human is still essential here

Humans remain responsible for choosing automation priorities, assessing risk, reviewing example responses and projected savings, and deciding when a category is safe to automate versus when human handling is still needed.

How people use this

Top ticket driver analysis

AI analyzes categorized support volumes and trends to highlight the issue types that create the most manual workload and are strongest candidates for automation.

Zendesk Explore / Salesforce Service Cloud

Self-service gap detection

AI reviews repeated ticket categories against existing help content to find questions that could be deflected with better articles or bot flows.

Intercom Fin / Zendesk AI

Automation candidate scoring

AI ranks support categories by frequency, resolution simplicity, and policy risk so teams can decide which workflows to automate first.

Forethought / Ada

Need Help Implementing AI in Your Organization?

I help companies navigate AI adoption -- from strategy to production. Whether you are building your first LLM-powered feature or scaling an agentic system, I can help you get it right.

LLM Orchestration

Design and build LLM-powered products and agentic systems

AI Strategy

Go from idea to production with a clear implementation roadmap

Compliance & Safety

Build AI with human-in-the-loop in regulated environments

Related Prompts (1)

Latest community stories (2)

News
News

What's new in Zendesk: May 2026

A new automation potential report analyzes your customer conversations and identifies requests that can be automated with AI agents. This report provides brand-specific insights and sample ticket data, showing you exactly how an AI agent would respond to customer inquiries.

(Advanced) Agentic AI for advanced email AI agents is now generally available. This enables AI agents to understand emails, answer questions, automate procedures, and escalate when needed, reducing back-and-forth with customers.


Copilot now includes macro content suggestions and trust and safety recommendations. These new recommendation types help you improve agent productivity and account security without spending time going through complex settings.


Generative search in help center now supports a follow-up question for customers using the Web Widget. This enhancement creates a smoother path from self-service search to a conversational experience with an AI agent, reducing repetition and keeping context from the original search.

CR
Colleen RomeroZendesk Documentation Team
May 1, 2026
LinkedIn

Most AI agents in customer support are just chatbots with fancy wrappers.

Most AI agents in customer support are just chatbots with fancy wrappers.

And I hate to break it to you, but they are the exact wrong place to start.


Everyone wants to jump straight to automated responses. But you cannot improve what you cannot clearly see.


Here's what I mean - we worked with an EV client drowning in support tickets. Humans were trying to sort requests manually across 12 main categories and over 60 subcategories.


The result? Inaccurate data, untrusted metrics, and total operational chaos. They didn't need a chatbot - they needed an AI agent to clean their data.


So we deployed an AI agent strictly for categorization. No complex reasoning. No generating responses. Just structuring the incoming requests.


The reality is:

→ Categorization is a safe, low-overhead starting point

→ It doesn't require massive models - open-source works perfectly

→ It reveals exactly which high-volume tickets you should actually automate


Only when you have pristine data do you start automating responses. Because that is when you clearly know where the biggest value for automation lies.


Stop chasing the generative AI hype. Start structuring your data.


If you are building AI into your operations, what is your first step? Are you categorizing your data first, or jumping straight to chatbots?


#AI #Automation #AIAgents #CustomerSupport #TechLeadership

EV
Eugene VyborovCEO at Ability.ai
Mar 19, 2026
Identifying which support issues to automate - People Use AI