Customers love to talk. The proliferation of LLMs such as ChatGPT, Claude, Gemini, and Perplexity, has fuelled demand for experiences that begin with natural language interactions, rather than searches for key terms or brand names. The additional of realistic voice capabilities has driven the trend further and it's drastically changing CX.
Figures from Adobe show that in the first quarter of 2025, retail sites in the US saw a 393 percent increase in traffic from AI sources. In 2024, Adobe reported a 1,300 percent spike in traffic from AI sources.
The trend is also evident in CX Network's annual research into the state of CX.
When 342 practitioners were asked to select the three CX trends they believe will influence their role to 2030, the top five most selected responses included: agentic AI/AI agents (selected by 21 percent), AI-first customer journeys/ customers using AI for product research (selected by 17 percent), and conversational AI chatbots and virtual assistants (selected by 16.5 percent).
The research also confirmed that AI-first customer journeys are not just a top CX trend, but also an emerging customer behavior. It emerged as the third most popular response, selected by 29 percent and followed by awareness of how AI works / uses customer data (36 percent), and demand for convenience (30 percent).
Furthermore, 15 percent of respondents said they plan to invest in conversational AI chatbots and virtual assistants this year to achieve their strategic CX goals.
Conversational AI has come a long way since rules-based automations repeatedly told customers their queries couldn't be addressed. Now powered by LLMs and agentic AI, written and spoken conversational experiences are now changing how customers and organizations engage. Today's technology makes it even easier for customers to interact using natural language and without the prior constraints of bots that struggle to interpret colloquial speech patterns, muddled intent, or even strong accents.
Through five cutting-edge conversational AI use cases, this article demonstrates how natural language is changing CX.
1. Klarna: Scaling voice AI, with human handoff
Klarna, one of the world's largest buy-now-pay-later providers, is no stranger to ambitious automation projects – or hedging its bets on AI. In 2025, while preparing for its IPO, it made headlines for reporting its AI assistant had taken on the job of 700 human customer service agents, and as a result, it was reducing hiring and laying off human employees.
However, the move didn't work out – service quality stalled and customers started to complain. Klarna backtracked and replaced its service team with humans.
Since then, the quality of AI has improved drastically and organizations have a much deeper understanding of where it should – and should not – be deployed. As a result, Klarna now has one of the leading use cases on how voice AI can be used to enhance human support, while still offering the option of a live agent.
Because a large share of Klarna's incoming calls – which traditionally required waiting for a human agent – were informational, Klarna deployed voice AI as the first line of phone support for its US-based customers. The voice AI was trained to deal with queries about payment statuses, product guidance, or next steps.
Built on ElevenAgents it was capable of handling customer requests through low-latency, natural dialogue, but crucially it also hands over to human agents when required. The result was a 10x decrease in resolution time for 35 million customers, reduced phone queues, and human agents freed to focus on complex cases.
2. Deliveroo: Three course conversational AI to support operations
Not all voice AI use cases are customer facing. While some organizations are using voice AI to enhance onboarding, others are using it to enhance back-end operations.
Deliveroo is one of Europe's largest on-demand delivery networks with more than 176,000 local partner businesses and thousands of deliver drivers. It deployed ElevenAgents across three use cases:
- Automatically contact rider applicants who had been inactive for more than two weeks. The voice agent confirmed the applicant's continued interest and guided them through the next steps to help them complete onboarding in their preferred language.
- Automatically confirm the live status of restaurants flagged as potentially closed. The agent placed outbound calls, validated opening hours, and updated Deliveroo's systems in real time.
- Proactively contact sites that had not activated tags after five business days of delivery, with the goal of guiding managers through installation and activation steps. In early results, Deliveroo's ElevenLabs Agent was able to successfully contact 86 percent of partner sites, helping to drive meaningful increases in activation.
Each use case runs as an outbound voice agent in the rider or partner's preferred language.
The voice agents reached more than 80 percent of target riders, verified 75 percent of flagged restaurants by phone, and contacted 86 percent of partner sites for tag activation. What previously took days or weeks of manual outreach can now happen in hours.
3. Macy's: The virtual shopping assistant that powered a 400% spending surge
AI traffic shows stronger conversion rates. Adobe's research confirmed that shoppers who use conversational AI to assist their shopping journeys, convert better than those whose journeys begin with paid search or email marketing.
Demonstrating the point, when US department store Macy's introduced its Ask Macy's shopping assistant, it saw those who engaged with it spend up to 400 percent more than shoppers who were not using the tool.
Ask Macy's helps customers find products faster by offering personalized recommendations. Shoppers who engage with the bot are also encouraged to explore complementary products – "complete the look" suggestions or curated bundles – which has been shown to significantly increase basket size and overall spend.
4. Raph Lauren: Ask Ralph brings in-store stylist experience to mobiles
When it comes to pioneering digital CX in fashion, Ralph Lauren has long led the pack. It was one of the first luxury brands to sell online and has since developed the PoloTech "smart shirt" that can track biometric data.
Today, it is pioneering an in-the-hand luxury omnichannel experience with its Ask Ralph conversational AI shopping assistant.
Ask Ralph brings the in-store stylist experience to customers' mobiles, allowing customers to receive personally curated outfit recommendations from the Polo Ralph Lauren collection by typing natural language prompts such as "What should I wear to a wedding in summer?" or "How can I style a pair of pink ballet pumps?"
It is built on Microsoft's Azure OpenAI platform, meaning the assistant can pull live from inventory and the brand's extensive visual archives to ensure recommendations remain up-to-date and on brand. Ask Ralph is available in the US to Apple and Android users and marks a significant step in the luxury brand's digital customer experience (CX) strategy.
5. Malaysia Airlines: Multi-lingual Mavis helps customers with time-sensitive inquiries
At Malaysian Airlines, service agent Mavis is supporting customers across critical moments of the travel journey – and when the query is too complex or sensitive for AI to handle, it brings a human into the conversation.
Mavis can engage travelers 24/7 to address high-value, time-sensitive inquiries, including flight status and schedules, booking and itinerary details, check-in access, boarding gate information, lowest fare discovery, seat upgrades, and more.
It supports travelers in Malay and English with additional languages planned to better serve a global customer base.
Built on Ada – which created the Agentic Customer Experience (ACX) operating model – Mavis reflects Malaysia Airlines' broader omnichannel vision, spanning web and mobile experiences. Currently, expansion plans include bringing AI-powered support into new channels and use cases, including voice and agent-assist capabilities, as well as an itinerary builder.
The last word on conversational AI
As conversational AI matures with stronger guardrails, more reliable orchestration, and enterprise-grade governance, organizations are moving beyond single-use-case pilots to multi-function deployments from a single platform.
As customers increasingly demand human-like, natural language experiences, this technology is making it even easier for them to interact without the prior constraints of bots that struggle to interpret colloquial speech patterns, muddled intent, or even strong accents.
As standards rise, organizations must understand and harness the power of a new generation of AI tools that can streamline CX, while also delivering benefits to employees and the wider organization.
Find out more about conversation AI in the
CX Network report, The new rules of conversational AI
Quick links
- Why voice fraud is CX's most underestimated problem
- How the WNBA is redefining fan engagement through community-led experiences
- Reinventing search and discovery within a digital ecosystem