In recent years, businesses of all sizes are racing to adopt artificial intelligence (AI) agents and chatbots to serve their customers — especially when customers reach out for support. The promise is clear: faster responses, 24/7 availability, lower cost per interaction.
However — there’s a subtle but critical mistake being made: many organisations focus on the flashy AI front-end before securing the underlying data that powers it. The result? AI tools that produce inconsistent answers, escalate too many cases to human agents, or worse, damage brand trust.
The Real Foundation: Knowledge Base + Ticket Data
In recent years, businesses of all sizes are racing to adopt artificial intelligence (AI) agents and chatbots to serve their customers — especially when customers reach out for support. The promise is clear: faster responses, 24/7 availability, lower cost per interaction.
However — there’s a subtle but critical mistake being made: many organisations focus on the flashy AI front-end before securing the underlying data that powers it. The result? AI tools that produce inconsistent answers, escalate too many cases to human agents, or worse, damage brand trust.
Here’s the truth: for an AI agent (or chatbot) to perform well, it needs good, reliable input data. Two key sources:
- Knowledge Base (Help Center)
- This is the repository of documented articles, FAQs, guides, troubleshooting steps, internal & external content, multilingual versions, etc.
- It serves both your customers directly (self-service) and your support agents (internal reference).
- This is often neglected or under-resourced.
- Without a well-maintained knowledge base, AI agents have nothing reliable to draw from.
- Example stats: 69% of consumers first try to solve their issue themselves — yet less than one‐third of companies offer self-service options such as a knowledge base.
- Businesses that use knowledge bases have been shown to get 23% fewer customer support tickets.
- Customer Service / Ticket Data
- All the historic tickets, chat logs, emails, incident resolutions, knowledge of how agents solved things — this is “real world” data.
- AI agents benefit hugely from this: the patterns, the language, the resolutions.
- If you deploy AI without incorporating ticket data, you risk angering customers because the AI doesn’t reflect your business context.
Why Build the Foundation First?
- According to one survey: 61% of customers prefer using self-service (knowledge base) for simple issues rather than contacting a live agent.
- Another: 81% of customers want more self-service options so they can find answers on their own.
- From the AI side: the market for AI in customer service is projected to reach almost US$47.8 billion by 2030.
- Yet the same source notes that 61% of companies report their data assets (for AI) aren’t ready.
All of this points to a key message: you can deploy an AI agent — but if the underlying knowledge base and ticket data aren’t solid, the ROI will suffer and customer frustration will rise.
Practical Steps to Prepare Your Data and Systems
Here’s a roadmap for making your organisation AI-ready in customer service:
- Establish / Clean Up Your Knowledge Base
- Create or refine your Help Centre: articles for customers, internal articles for agents.
- Categorise content: internal vs external; languages (e.g., English, Thai, Chinese) if you serve multilingual customers.
- Make sure content is searchable, up-to‐date, clear and structured.
- Monitor usage: what articles are frequently used? Which search terms return no results? Use that insight to fill gaps.
- Create or refine your Help Centre: articles for customers, internal articles for agents.
- Aggregate Ticket & Interaction Data
- Collect past tickets: what queries do customers ask? How did agents resolve them? What language/terminology used?
- Apply tags / categories so patterns emerge.
- Consider integrating chat logs, email threads, forum discussions.
- Cleanse and anonymise data if required (privacy / compliance).
- Collect past tickets: what queries do customers ask? How did agents resolve them? What language/terminology used?
- Connect Your Knowledge Base & Ticket Data as AI Input Sources
- When you deploy an AI agent / chatbot, ensure it can access both the knowledge base and your historical interaction data.
- Update the knowledge base continuously (because products change, processes change).
- Track feedback: when the AI fails or escalates, feed that back to update the knowledge base.
- When you deploy an AI agent / chatbot, ensure it can access both the knowledge base and your historical interaction data.
- Select a Customer Service Platform that Supports Data Readiness
- If you use a tool such as Zendesk, you’re in good shape: it offers modules for knowledge base, ticketing, analytics.
- These platforms make it easier to extract, clean and structure data for AI consumption.
- If you use a tool such as Zendesk, you’re in good shape: it offers modules for knowledge base, ticketing, analytics.
- Measure & Iterate
- Track metrics such as: number of tickets deflected by self-service, escalation rate from AI agent to human, CSAT for AI responses, reduction in average resolution time.
- Review quarterly: what knowledge base articles need updating? What ticket categories are trending?
- Track metrics such as: number of tickets deflected by self-service, escalation rate from AI agent to human, CSAT for AI responses, reduction in average resolution time.
The Long-Term Payoff
Yes — building a solid knowledge base and collecting service data takes effort. But the return is compelling:
- Reduced load on support agents → lower cost per interaction.
- Faster, more consistent responses to customers → improved satisfaction and retention.
- AI agents become more effective over time (learning from good data) → greater scalability.
Competitive advantage: while many businesses deploy AI, fewer invest in the data foundation. If you do, you’ll likely outperform peers.
How DEMETER ICT Can Help
At DEMETER ICT, we specialise in helping SMBs and enterprise-clients implement, configure and optimise Zendesk (and other customer-service platforms) for exactly this purpose. We assist with:
- Setup of the knowledge base / help centre (languages, internal vs external access)
- Migration and structuring of historical ticket data and interaction logs
- Integration and readiness for AI-powered customer service agents
- Ongoing governance: knowledge-base maintenance, data updates, analytics
👉 If your organisation is planning to deploy AI in your customer service, let’s talk. We’ll help you build the data foundation first — so that when your AI agent goes live, it delivers real value, not just hype.
About the author
Dr. Varanyu Suchivoraphanpong is currently the Founder & CEO of DEMETER ICT, a Premier Partner of Google and Zendesk in the APAC region. The company has the largest customer base for Google and Zendesk services in APAC including Greater China, with a combined total of more than 4,600 business customers. Dr. Varanyu has over 25 years of experience in information technology, having previously held senior executive positions in major organizations including banks, IT service providers, and business consulting firms.
