AI customer service is no longer just a chatbot added to a help centre. Done properly, it is a practical way to improve service quality, reduce repetitive work, and give customers faster answers without removing the human support they still need for complex issues.
That was the real story in SuccessCX’s webinar with RMS Cloud and Zendesk.
RMS Cloud is a global property management technology provider operating across 70 countries. Its support environment includes different regions, time zones, customer segments, product modules, and service-level agreements. In the webinar, Garth Kay, VP of Customer Support at RMS Cloud, explained how RMS moved from regional support teams to a global follow-the-sun model.
With SuccessCX and Zendesk, the team focused on data hygiene, knowledge management, self-service, skill-based routing, AI agents, and better support workflows.
The lesson is simple: AI in customer service works best when it is built on strong foundations. The tool matters. But the operating model matters more.
What is AI in customer service?
Customer service refers to the way a business helps customers before, during, and after they use a product or service.
AI in customer service means using artificial intelligence to assist, automate, route, summarise, analyse, or resolve parts of the customer support process.
That can include AI agents answering common questions, customer service AI tools suggesting responses to human agents, natural language processing detecting customer intent, or machine learning helping support teams identify trends in customer interactions, ticket volumes, and customer requests.
AI customer service in plain English
AI customer service is the use of AI tools to help customers get answers faster and help support teams work more efficiently.
It might look like:
AI agents answering simple customer questions
AI agents can respond to repeat customer questions, find relevant knowledge base articles, and provide immediate answers, much like well-designed customer service chatbots that balance automation with human support.
AI tools assisting human agents
Customer service AI can suggest replies, summarise support conversations, recommend macros, and surface past interactions, while also supporting agent training and ongoing coaching with AI in CX.
AI systems improving support operations
AI systems can identify trends, analyse customer sentiment, support intelligent routing, and help leaders understand where service quality is improving or falling behind, illustrating the broader rise of AI in customer service.
The point is not to replace every person in support. The point is to help the team give better answers, faster.
Why AI customer service is now a priority
Customer expectations have changed. Most customers now expect quick replies, accurate answers, and 24/7 access to support, especially for simple customer questions.
At the same time, many support teams are dealing with higher ticket volumes, tighter budgets, and more pressure to reduce operational costs. They need to improve customer satisfaction, but they cannot keep adding people to every backlog.
This is where AI customer service becomes useful.
The pressure on modern support teams
Customer service teams are not only chasing lower customer service costs. They are trying to address customer needs, deliver personalised support, maintain exceptional service, and do it with fewer resources.
AI in customer service can help by automating routine tasks across common customer service functions while lowering operational costs, especially when organisations choose the best AI customer service tools for their support strategy.
The key benefits of AI in customer service
The key benefits are practical.
AI can handle routine tasks, provide instant support, help human agents focus on higher value work, and improve agent efficiency by putting the right information in front of the right agent.
Faster support
AI agents can reply instantly to simple customer requests and reduce wait times.
Better agent efficiency
AI tools can give support agents response suggestions, summaries, and relevant customer data.
More consistent service quality
AI powered customer service can help teams give more consistent answers across regions, channels, and time zones.
Lower operational costs
By automating routine tasks and reducing repeat tickets, AI can reduce operational costs and help teams scale with fewer resources.
But AI in customer service only works when the business has clean data, accurate knowledge, clear escalation rules, and strong change management.
The RMS Cloud example
RMS Cloud did not start with the question, “How do we launch a bot?”
The better question was, “How do we create a support model that can scale globally without compromising customer experience?”
The challenge: global support with regional inconsistency
RMS Cloud had regional support teams working independently. Customers operating across regions could receive different experiences depending on where they were served.
Some customers had strict SLAs, and in some cases those SLAs were not measured in business hours. The business needed 24/7 support in certain situations.
The goal: scalable support without losing quality
The goal was to build a support model that could scale globally while still giving customers a consistent experience.
That meant improving ticket data, reviewing taxonomy, building a cleaner knowledge base, creating clearer SOPs, moving towards skill-based routing, and giving support agents better visibility into priorities and SLAs.
Why the foundation mattered
Without clean data, RMS Cloud could not properly understand what customers were asking.
Without a strong knowledge base, AI agents would not have reliable content to draw from.
Without clear routing, customer queries could not always reach the right agent quickly.
This is why the transformation began with operations before AI.
Self-service is not about blocking customers
One of the strongest points from the webinar was Garth’s view on self-service.
Self-service is not about stopping customers from reaching human support. It is about giving customers faster answers when the issue is simple enough to solve without waiting in a queue.
Good self-service feels like instant support.
The problem RMS Cloud needed to solve
For RMS Cloud, many customer requests were basic how-to questions. Customers were often bypassing help centre articles and logging tickets straight away.
That created more backlog, slower first response times, longer resolution times, and pressure on human customer service teams.
The role of the knowledge base
The answer was not just “add AI”.
The answer was to improve the knowledge base so AI agents could provide useful answers.
Why knowledge gaps slow everything down
Knowledge gaps create repeat tickets, inconsistent answers, and more pressure on human support.
If the same question keeps reaching human agents, the issue is often not just customer behaviour. It may be a content problem, a product education problem, or a knowledge base structure problem.
How AI agents support self-service
AI agents can help customers resolve low-complexity questions in real time. In RMS Cloud’s case, their advanced AI agent, Miles, was designed to sit in front of customers and guide them towards answers using the company’s knowledge base.
This is one of the most effective use cases for AI in customer service.
What AI agents should handle first
The best starting point is usually high-volume, low-risk customer questions.
That might include:
How-to questions
These are repeatable questions with a clear answer.
Account or access questions
These can often be solved through guided workflows.
Basic troubleshooting
AI agents can step customers through simple fixes before escalating.
Product navigation questions
AI can help customers find the right article, feature, or support path.
If a question is too complex, too sensitive, or too unclear, the AI should hand the issue to human support.
AI agents assisting customer service agents
RMS Cloud also used AI on the agent side, not just the customer-facing side, similar to how AI copilot capabilities inside tools like Zendesk assist agents directly in their workspace.
This is where customer service AI can have an immediate impact. AI tools can help customer service agents by surfacing knowledge base articles, past interactions, response suggestions, macros, and ticket context inside the agent workspace, which can significantly accelerate agent ramp-up and productivity.
Why agent-side AI is often the safest starting point
Agent-side AI lets teams test AI in customer service without putting every answer directly in front of customers.
It gives support agents help while keeping human review in the workflow.
Suggested replies
AI can draft or recommend responses based on context.
Ticket summaries
AI can summarise long support conversations so agents do not have to read every message from scratch.
Knowledge recommendations
AI can recommend knowledge base articles linked to the customer’s issue.
Next-step guidance
AI can suggest what the agent should check, ask, or do next.
For support agents, that means less time hunting for information and more time solving the issue.
Knowledge management is the hidden engine
Every AI powered customer service project eventually runs into the same truth: AI is only as useful as the knowledge it can access.
For RMS Cloud, knowledge gaps were part of the original problem. Customers were asking repeat questions. The team had deep knowledge, but the system was not always structured in a way that made those insights easy to report on or reuse.
Why AI needs clean content
AI in customer service becomes less about technology and more about operations once it starts using your own content.
If the knowledge base is out of date or vague, AI agents will struggle.
If the content is accurate, tagged, and written around real customer needs, AI powered solutions become far more useful.
Strong AI knowledge content should be clear
The article should answer one customer question properly.
Strong AI knowledge content should be structured
Headings, steps, tags, and categories help both people and AI systems find the right information.
Strong AI knowledge content should be maintained
AI systems require continuous monitoring and improvement for optimal performance. So does the knowledge base that supports them.
Intelligent routing and skill-based support
Another important shift for RMS Cloud was moving towards intelligent routing and a skill-based support model.
Before the transformation, support work was more regionally based. After improving data and taxonomy, RMS Cloud could see which product areas, modules, or issue types were driving volume.
That made it easier to route customer queries to the right agent.
How intelligent routing improves customer experience
Intelligent routing helps support teams match work to expertise.
Simple requests can be answered through self-service or AI agents. More complex issues can go to human agents with the right technical expertise. Urgent issues can be escalated faster.
High-risk customer interactions can be prioritised before they damage customer relationships.
Better routing improves speed
Customers get to the right person faster.
Better routing improves quality
Human agents can work on topics they know well.
Better routing improves team morale
Agents spend less time being bounced between unrelated issues.
Measuring customer sentiment and customer satisfaction
AI can also help support teams analyse customer sentiment and identify where service interactions are becoming risky.
Using natural language processing and machine learning, AI tools can detect frustration, urgency, satisfaction, confusion, or emotional intensity in customer messages.
How customer sentiment should be used
Sentiment analysis should not be treated as a gimmick.
It should help teams act faster when a customer is unhappy or when an issue is likely to escalate.
Route negative sentiment faster
Negative customer sentiment can be sent to prioritised escalation queues.
Coach agents with real examples
Managers can use AI insights to coach agents on tone, clarity, and empathy.
Analyse service trends
AI can analyse customer sentiment across support interactions and identify trends over time.
Customer sentiment analysis is especially useful when combined with CSAT, first contact resolution, reopen rates, and resolution time, because together they show how AI is improving overall customer satisfaction.
Together, these metrics help leaders understand whether AI in customer service is improving the customer experience or just moving tickets around.
From cost centre to value driver
Customer support is often treated as a cost center. That is a mistake.
When support teams have the right customer data, they can identify trends, reduce churn risk, improve product feedback loops, and surface expansion opportunities.
How AI can support revenue and retention
AI in customer service can help by analyzing customer data, spotting customer behavior patterns, and showing where customers need training, onboarding, or proactive outreach.
This is where transforming customer service becomes more than cost reduction and connects directly to using AI to improve customer experience across channels.
Yes, AI can help reduce customer service costs and lower operational costs. But the bigger opportunity is better retention, stronger customer relationships, and personalised support that feels timely rather than forced.
Support can reveal churn risk
Repeated complaints, negative sentiment, and reopen rates can point to accounts that need attention.
Support can reveal product gaps
High-volume customer questions can show where the product or onboarding experience needs improvement.
Support can reveal expansion opportunities
Support conversations can show when customers are ready for new features, integrations, or services.
Forecasting and capacity planning
AI can also improve workforce planning.
Support leaders need to know when ticket volumes are likely to rise, which issue types will drive the increase, and where staffing pressure will appear, often prompting investment in modern contact centre solutions that leverage AI.
Using historical data to plan ahead
Historical data can help forecast peaks and troughs, especially when combined with product launches, seasonal patterns, customer segments, and known incidents.
For RMS Cloud, global support meant thinking beyond a single region or standard business hours.
For similar organisations, AI in customer service can support forecasting, capacity planning, and follow-the-sun operations with fewer resources.
Forecast demand by region
Global teams need to understand where support load is increasing.
Forecast demand by issue type
Product, billing, onboarding, and technical issues may need different staffing plans.
Forecast demand by channel
Chat, email, phone, and self-service all create different operational demands.
Call management and voice AI
Call management is another area where customer service AI is developing quickly, especially as contact centre AI reshapes how phone-based support operates.
Voice AI can help with conversational IVR, real-time transcription, after-call summaries, call reason tagging, and escalation prompts, extending the impact of modern IVR solutions in call centres.
Where voice AI can help
Voice AI is especially useful where calls are high-volume, repetitive, or hard to analyse manually, making robotic call handling the next step in automation.
Real-time transcription
Calls can be captured and summarised without agents writing everything manually.
After-call summaries
AI can reduce admin time after each call.
Escalation prompts
AI can detect urgency or frustration and suggest escalation.
QA and coaching
AI can help managers review more calls and coach agents with better evidence.
How to implement AI for customer service
Implementing AI should not start with a product demo.
Implementing AI well means connecting the technology to customer expectations, service interactions, governance, and real operational goals.
Start with an AI readiness review
Before choosing AI tools, leaders should ask:
- What customer questions drive the most volume?
- Which routine tasks slow agents down?
- Where are the biggest knowledge gaps?
- Which customer interactions need human support?
- What customer data is available, accurate, and safe to use?
- What does a good escalation look like?
- What does success look like after 30, 60, and 90 days?
Choose AI tools based on the operating model
A strong customer service solution should improve customer interactions and the customer experience, not just deflect tickets.
Implementing AI is a service design project, not just a software rollout.
Check help desk integration
Your AI tools should work with your existing customer service software.
Check CRM integration
AI is stronger when it can access the right customer context safely.
Check reporting
You need to measure whether AI is improving customer satisfaction, resolution time, and service quality.
Check governance
AI systems need access controls, escalation rules, audit trails, and security measures.
Start with high-volume, low-risk use cases
The best starting point is usually high-volume, low-complexity work.
That could include password resets, account access questions, simple how-to guides, billing FAQs, booking questions, product setup steps, or basic troubleshooting.
Why “perfect” is the wrong target
Garth’s advice from the webinar was direct: do not aim for perfect.
Start with use cases where the knowledge is clear, the risk is manageable, and the value is obvious.
Pick one clear use case
Do not try to automate every customer journey at once.
Define success before launch
Decide what good looks like before the pilot begins.
Review real conversations
Use real support conversations to test whether AI agents are helping or creating friction.
Keep human support in the loop
Successful AI adoption in customer service requires a hybrid approach with human support.
There should always be clear rules for when AI hands over to a person.
When AI should escalate to human agents
AI should escalate when the issue is complex, sensitive, urgent, commercial, or emotionally charged.
Common triggers include:
Angry or frustrated customer sentiment
Negative sentiment should be handled carefully.
Repeated failed answers
If AI cannot help after a defined number of attempts, it should stop trying.
Refunds, security, and account risk
These areas need stricter controls.
Enterprise or high-value customers
Some customer relationships need more direct human support.
RMS Cloud did this well. Their AI champions tested Miles using real support conversations, checked whether the responses matched brand expectations, and helped identify where content needed improvement.
Set guardrails before going live
AI systems require continuous monitoring and improvement for optimal performance.
That means every implementation needs guardrails.
What your AI guardrails should cover
Your governance model should define what AI can answer, what AI must not answer, when AI should escalate, which systems AI can access, what customer data can be used, what tone the AI should use, how outputs will be reviewed, and what security measures are required.
Accuracy
The answer should be correct and based on approved knowledge.
Brand tone
The AI should sound like the business, not a generic bot.
Escalation
Customers should not get trapped in endless loops.
Privacy and access
AI should only access the customer data it genuinely needs.
Measure the right things
A common mistake is measuring AI in customer service only by deflection.
Deflection is useful, but it is not enough.
Metrics that matter
The best metrics include CSAT, bot satisfaction, first contact resolution, reopen rate, average handle time, first response time, full resolution time, cost per contact, escalation rate, knowledge gap rate, and agent efficiency.
Customer satisfaction
Customer satisfaction shows whether customers are getting a better experience.
First contact resolution
This shows whether the issue is being solved properly the first time.
Reopen rate
A low reopen rate suggests customers are getting useful answers.
Resolution time
This shows whether AI is helping teams resolve cases faster.
Cost per contact
This helps track whether AI is lowering operational costs.
Some benchmarks suggest mature AI adopters report a 38% lower inbound call handling time, while AI can reduce average case handling time by 35% in some environments.
Other AI business cases model a reduction in operational costs by up to 30%, agent efficiency increases of around 33%, and customer satisfaction improvements of up to 15% through personalisation.
Treat those numbers as planning benchmarks, not promises. Your result depends on your data, content, workflows, customer needs, and implementation quality.
What AI in customer service can realistically automate
AI can automate over 70% of customer queries in some support environments, especially when the business has a strong knowledge base and many repeat questions.
AI-powered chatbots can also resolve up to 50% of support questions instantly when the scope is clear and the content is strong.
Best-fit use cases for automation
AI is best suited to routine tasks, repetitive customer questions, ticket triage, response suggestions, summaries, knowledge retrieval, sentiment analysis, and simple workflow automation.
Good automation candidates
These include simple FAQs, repeat how-to questions, order updates, password help, booking changes, basic troubleshooting, and article recommendations.
Poor automation candidates
Human support is still better for complex complaints, negotiation, sensitive accounts, unclear context, and moments where empathy matters.
Which AI is best for customer service?
The best AI for customer service is the one that fits your existing customer service software, workflows, knowledge base, CRM, reporting needs, and governance requirements.
Why Zendesk AI makes sense for Zendesk customers
For Zendesk customers, Zendesk AI is often the logical starting point because it works inside the service environment your teams already use.
That reduces friction, speeds up adoption, and keeps AI closer to your actual support data.
The best AI is not always the flashiest AI
The wrong way to choose a customer service solution is to pick the tool with the flashiest demo.
The right way is to map AI for customer service against your real support operations.
What is an example of customer service AI?
A practical example of customer service AI is an AI agent that answers how-to questions using your help centre content, then escalates complex queries to human agents.
RMS Cloud’s AI agent example
RMS Cloud’s Miles is a good example.
It was launched as an advanced AI agent to help customers get timely answers, while the team monitored conversations, reviewed quality, and continued improving the knowledge base.
Why the example matters
The important part is not just that RMS Cloud launched an AI agent.
The important part is that they involved their support teams, tested real customer queries, checked the quality of responses, and used feedback to improve the system.
What is the 30% rule in AI?
There is no universal “30% rule” in AI.
In customer service, people often use 30% as a rough planning benchmark for potential efficiency gains or cost reduction.
How to think about the 30% benchmark
For example, AI can reduce operational costs by up to 30% in some environments when it is used to automate routine inquiries, improve routing, reduce handle time, and help agents work faster.
Do not treat 30% as guaranteed
A 30% improvement depends on your current state.
If your knowledge base is weak, ticket data is poor, or the team does not adopt the system, the result will be lower.
Who are the big 4 AI agents?
There is no official “big 4 AI agents” list for customer service.
Most buyers compare AI agents based on the platform they already use, such as Zendesk, Salesforce, Intercom, Freshworks, ServiceNow, Genesys, or other customer service software.
The better buying question
For a Zendesk environment, the more useful question is not “who are the big 4?”
It is “which AI agents can work with our knowledge base, workflows, reporting, security, and support model without creating a mess?”
Platform fit matters
AI should fit the way your support operations actually work.
Governance matters
AI should be safe, auditable, and easy to improve.
Adoption matters
If agents do not trust the tool, it will not deliver the result you want.
The real takeaway
AI in customer service is not a shortcut around good operations.
It is a multiplier.
If your knowledge base is messy, AI will expose it. If your ticket data is poor, AI will struggle to prioritise. If your escalation rules are unclear, customers may get stuck. If your team does not trust the system, adoption will stall.
What RMS Cloud shows us
The RMS Cloud transformation was not “we switched on AI and everything changed.”
It was cleaner data, better knowledge, better routing, better workflows, stronger governance, and then AI layered on top.
That is how AI powered customer service becomes more than a chatbot.
It becomes a better way to run support.
The practical lesson
- Start with the support model.
- Fix the knowledge.
- Clean the data.
- Involve the team.
- Pilot with low-risk questions.
- Measure the customer experience.
- Then scale what works.
Ready to explore AI customer service?
If you are already using Zendesk and want to understand where AI could make the biggest difference in your support operations, SuccessCX can help you assess the right starting point.
It is a practical review of your current support model, your knowledge base, your customer requests, your ticket data, and your highest-volume use cases.
From there, you can build a phased roadmap for AI for customer service that improves the customer experience without overwhelming your team.
