Ask the people who run customer experience what AI has changed, and few of them lead with speed. They lead with memory. What customers notice first, they say, is that they no longer have to start over. The complaint raised last week is still on file when they call back, and the preferences they gave once are already in place. NervNow convened seven leaders across travel, insurance, retail, food and consumer appliances to talk about what that has fixed for customers, and what AI still cannot do for them.

For Pranshu Rastogi, chief experience officer at The Sleep Company, the change is continuity. Every interaction used to feel like a fresh start, with customers repeating their story across channels. Now the context carries over. The customer, in his words, feels known rather than processed, and the practical payoff is that people spend less time explaining and more time getting to a solution.

Abhishek Mago, CMO at the travel marketplace Cheapfaremart, calls the same thing contextual continuity. Journeys used to be fragmented, he said: a campaign, a support call, an abandoned cart, a return weeks later, each handled in isolation. AI ties those moments together and reads intent across them. For him the measure of success is effort. “Customers don’t measure success by how quickly a company processes information,” he said. “They measure it by how effortlessly they achieve their goal.”

Shubham Choudhary, an SVP and business head at Policybazaar, describes it as relevance at scale. Every message his team sends, whether a renewal reminder or a nudge to re-engage, is shaped by what the customer has bought, said and is likely to need next. The response data bears it out. “When customers feel a message was written for them, they engage,” he said. “When they feel like a number, they leave.”

Udbhav Mishra, who leads consumer strategy and experience at the bus operator FreshBus, points to listening. Putting speech AI across the contact centre let his team hear every conversation instead of a sample, which changed both what they chose to fix and how quickly they fixed it. He is careful, though, about who gets the credit for the fix itself.

The complaints AI has quieted were rarely about the product. Venkat Ramanan, who heads customer experience and digital transformation at the appliances brand Lifelong, said his customers used to call mostly to ask where things stood: when a technician would visit, when a replacement would ship, when a refund would clear. Those calls arrive in waves, choke a contact centre and push up abandoned-call rates, and the same people keep dialling back. An AI agent now answers the status questions directly, and when a service deadline is about to slip, it raises an internal alarm so a person can step in. First-contact resolution has climbed, and escalations get caught before they boil over.

Pranshu Rastogi framed the same era around effort. Customers do not mind problems, he said, they mind having to chase the fix: the long waits, the repeated information, the handoff from one agent to the next. AI cut that friction by putting answers within reach at once and moving people to a resolution faster.

At FreshBus, the problem was silence. Udbhav Mishra said a delay rarely upset customers on its own. What upset them was hearing nothing while it played out. Once real-time updates went out automatically along the journey, complaint volumes fell sharply, even on trips where nothing about the delay had changed. “The issue often had not changed,” he said. “The anxiety around it had.”

Ashish Bajaj, cofounder of the food brand 10on10 Foods, had run into the same gap from the build side. Holding one continuous view of the customer was a hard problem for years, he said, and stitching it together through a connected CRM and other workarounds only carried the team so far.

On how a company with no AI should start, the panel was close to unanimous, and none of them said “pick a tool.” Shubham Choudhary said most businesses do precisely that, deploying a chatbot that solves the wrong problem before anyone has mapped what the customer actually goes through. His advice is to find the single biggest pain point, work out how to solve it for one customer, then run that solution at scale. Take the highest-volume interaction, where a one percent improvement is worth a great deal, and start there.

Pranshu Rastogi put it as an order of operations: start with the problem, then the technology. The mistake he sees most is teams adopting AI because everyone else is, and then aiming it at a process that was already broken.

Dharmendra Raghuvanshi, CTO of Chai Point, approaches the question by deciding where AI should and should not go. He aims it at the invisible work, the forecasting and inventory and replenishment and routing, so the cup ends up hot, on time and exactly what was ordered. Every manual step in that chain is a place something can break, he said, so the job is to find each one, model it and let the system carry it.

A common worry is that AI for customer experience is built for companies with deep pockets, and the panel pushed back on it. Abhishek Mago called AI one of the biggest levellers in business. What used to need an enterprise budget is now within reach of a startup, and what separates the winners is how well a company understands its customers, which a small firm can manage as well as a large one. The trap, he said, is assuming AI demands a sweeping transformation. The businesses that do best start small, solve one specific problem and build from there.

Shubham Choudhary agreed that a good experience was never a function of company size. WhatsApp Business API, off-the-shelf CRM platforms and chatbots built on large language models are all available without an enterprise budget, he said, and a small business today can reach for capabilities that would have cost crores five years ago.

Venkat Ramanan made the cost case in plain terms. A generative-AI chatbot is not expensive to put in place, and the spend should be weighed against the calls it takes off human agents and the retention it buys back. His one caution for smaller firms is to choose a reliable services partner to build the thing collaboratively, rather than a vendor selling a finished product off the shelf.

Ashish Bajaj offered a rule of thumb for early-stage companies: budget roughly two to five percent of monthly GMV for tools, more in some sectors. Availability, he said, is not the constraint it used to be.

If the old scorecard was clicks and conversions, the panel has moved past it. Dharmendra Raghuvanshi tracks one number above the rest at Chai Point: order-to-serve time, the gap between someone tapping order and the cup reaching their hand. Demand forecasting lets the team pre-position stock and staff ahead of the rush, so the bottlenecks that used to show up at peak hours mostly do not. The effect he cares about is what happens after that.

Abhishek Mago has rewritten the metrics at the marketplace level. His team is moving past clicks, impressions and conversions as standalone measures toward customer effort, task completion, repeat interactions and lifetime value. What he wants to know is whether the company saw a need coming and met it before the customer had to ask.

Venkat Ramanan watches first-contact resolution for much the same reason. When the AI agent settles a query on the first call and flags the ones it cannot, the number that moves is how often a customer has to come back at all.

If every rival can license the same foundation models, where does an edge come from? Udbhav Mishra is blunt about it. The model is not the advantage. At FreshBus, what set the experience apart was two years of proprietary interaction data, the specific context of its customers and the emotional design built around the AI, none of which a competitor can stand up in a hurry.

Abhishek Mago lands in the same place from the brand side. As the technology levels out, he said, the difference comes from how well a company knows its customers, remembers their preferences and connects their experiences across touchpoints.

Voice AI is the use case everyone seems to be reaching for, and the panel was more cautious than the hype around it. Venkat Ramanan said the cases that justify it are narrow: high call volumes, heavily transactional conversations, the same information repeated again and again. His team has watched the demos and is holding back until the fit is right.

Udbhav Mishra has been deeper into it, and his read is sober. The demos all impress, he said, and then deployment brings you down to earth. At FreshBus most of the work goes into the gap between lab accuracy and the real world, where accents and emotional tone live.

Abhishek Mago saw the same gap in travel. His team explored voice AI for flight changes, booking confirmations and disruption handling, and the demos were impressive, but live use ran into accents, multilingual conversations, context that had to be held across turns and the raw emotion of a delay or a cancellation. The technology is close, in his telling. The last mile is where the effort goes.

The agreement broke down, usefully, on the question of limits. Udbhav Mishra drew the clearest line. Real-time communication during a delay had cut FreshBus’s inbound calls, because what most people wanted was clarity, and a system can provide that. Some moments are different. A cancellation the night before a wedding, an elderly traveller stranded somewhere unfamiliar: no bot closes that gap, he said. “AI handles disruption well,” he added, “until the customer stops wanting information and wants to feel heard.” The work now is getting the handoff right the moment emotion enters.

Venkat Ramanan put the same boundary in terms of complexity. AI does well on efficiency and on simple, repeatable problems, he said, but as a case gets more tangled the agent starts to flounder, and an aggrieved customer wants counselling and reassurance from a person.

Shubham Choudhary sees the same line inside insurance, where a wrong explanation can do real harm. AI has made customers better informed and quicker to ask the right questions, he said, but a complex product carries a thin line between explaining and misleading. When the AI cannot move the conversation forward, when a query is too specific for the data behind it, or when the customer simply asks for a person, that is the cue to hand to a human.

Dharmendra Raghuvanshi has built that boundary into how Chai Point runs. The automation sits on the operational side so the storefront stays calm, and the human moments are left alone. Tea is close to a ritual in India, he said, so the warmth in how it is served stays human by design.

Asked where they would point AI next, two of them named bets that have little to do with chatbots. Ashish Bajaj sees it reshaping health and nutrition, a category full of informed and skeptical customers. Generic advice is the problem, he said. When a device reads your vitals and tracks your steps, sleep and calories, the plan stops being built for an average person and starts being built for you, and that is when trust and follow-through go up. An AI frontline nutrition advisor is coming, in his view, and he is not hedging about it.

Dharmendra Raghuvanshi points to a use case most companies overlook, predictive maintenance. Everyone chases the customer-facing AI, he said, but in Chai Point’s brewing-machine business every unit is a revenue node, so catching a failure before it happens matters more than it looks. Live data reads the early warning signs and triggers a service visit before the customer ever notices the machine is struggling. It is not glamorous, which is part of why it stays underrated, but in a hardware-heavy operation uptime drives almost everything.

None of this leaves the customer experience job untouched. Ashish Bajaj was blunt about it: the routine work goes. He compared it to the bank ledger that the spreadsheet replaced, which forced the people doing it to learn new skills. Data crunching and quality checks are already thinning out, and the skill that matters now is directing that work through AI.

Pranshu Rastogi agreed that repetitive roles will shrink, and said the job is moving toward the parts that are hardest to automate. People will spend less time hunting for information and more on complex problems and human judgment, with AI taking the routine and people keeping empathy and the decisions that carry weight.

Abhishek Mago sees the role evolving. The future professional answers fewer questions and spends more time reading the insights AI surfaces and improving the moments that decide how a customer feels. The organisations that come out ahead, he said, are the ones that pair AI’s scale with human judgment and emotional intelligence.

Dharmendra Raghuvanshi kept returning to a question of where the technology belongs. Every minute his staff is not spending on stock counts or reconciliations is a minute back with the customer, and a good cup with a real smile is still, as he puts it, the product. He is also wary of what optimisation can cost. AI does whatever you point it at, he said, so a system aimed only at conversion will erode customer choice without anyone intending it to. The discipline is in choosing what to optimise for, and that accountability stays with the people building the system.