According to our latest report, 15 AI Trends in CX 2025, businesses are at a crossroads in their AI journey. 2024 saw all the headlines about AI breakthroughs from the likes of OpenAI, Google and Apple, but the reality on the ground is different. Leaders can often think AI is easy – a magic bullet. One AI vendor told us that out of 75 AI pilots run by a single prospect in 2024, only 3 made it to production – a sobering reminder that implementing AI is much harder than most expect.
So, what’s really happening with AI in customer experience? What should businesses focus on in 2025?
Here are the seven insights from the report to help you get moving.
1. The End of the magic bullet mindset
Leadership teams realise that AI isn’t the plug-and-play solution many thought it was. Successful AI implementation still requires robust project management, governance, design and testing processes – perhaps even more due to AI’s non-deterministic nature. One-prompt, request-response solutions might achieve 80% accuracy in prototypes, but to get to production-grade performance requires sophisticated infrastructure and solution design.
2. Holistic strategy is key
Organisations need to move beyond point solutions for problems like Retrieval Augmented Generation chatbots and develop comprehensive channel strategies. This means mapping customer intents, identifying preferred resolution paths, and thoughtfully triaging contacts across channels. The most successful companies are thinking bigger than just knowledge retrieval chatbots.
3. Introduce service design
Simply automating broken processes with AI won’t deliver real transformative results. It’s like slapping lipstick on a pig. You can make an effort to make it look pretty, but it is still a pig at the end of the day. Instead, we advocate for AI service design – a holistic approach to redesigning end-to-end experiences rather than just digitising existing workflows. This involves thorough user research, process mapping, and reimagining the entire customer journey.
4. Analytics evolution
According to Everest Group, 73% of Chief Information Officers report unclear success metrics as a challenge. We’ve observed similar trends this year both from enterprises and vendors. Trying to prove the value of gen AI is like trying to prove the value of the thread in your shirt. To measure the impact of AI, we recommend a three-tiered analytics framework:
- Interaction Level: Evaluating the immediate effectiveness of customer interactions.
- Journey Level: Tracking experience across multiple touchpoints.
- Business Level: Linking AI performance to organisational goals like cost savings and revenue impact.
5. ROI challenge
Organisations struggle with budget forecasting and ROI calculations as AI moves from pilots to production. Leadership must understand this is a long-term investment, with initial heavy investment before you can scale. Moreover, the frequent updates to generative AI models and changing cost structures make traditional ROI calculations particularly challenging.
6. Breaking down silos
The divide between AI and traditional NLU teams creates inefficiencies and drives up operational costs. Forward-thinking organisations are working to merge these teams to improve collaboration, speed up experimentation, and accelerate deployment. Crucially, this integration helps operational teams develop the skills they’ll need as AI becomes more central to customer experience delivery.
7. Bonus: The promise of agentic AI in customer experience
While in 2024, we saw the beginning of hype around agentic AI in customer experience, the reality is more nuanced. There is a fundamental divide in how the industry defines agentic AI. For AI purists, agentic AI represents end-to-end generative AI applications that autonomously make decisions and complete tasks. However, some vendors are simply rebranding pre-programmed, rule-based workflow logic as agentic.
2025 will likely continue to see more confusion than results, with AI agents deployed in areas where traditional rules might perform better and well-performing rule-based systems being mislabeled as ‘agentic.’ However, the technology shows promise for tackling complex, long-tail use cases that currently require human intelligence. Keeping an eye on how things develop will be important in the year ahead as technology inevitably improves.
Navigate the CX landscape with our trends report
These insights represent the beginning of our 15 AI Trends in CX 2025. The report is a must-read for organisations navigating this complex landscape. It provides valuable, real-world insights drawn from the last 12 months of hands-on experience at VUX Consulting, working with enterprises deploying the latest AI technologies.
Want to learn about all 15 trends and get detailed recommendations for 2025? Download our complete 15 AI Trends in CX 2025 report.
