March 23, 2026

In this episode of Stare Down the Bull, host Susan Hunt interviews technology and sales leader Tim Walsh about the long arc of automation and AI in customer experience, from early voice recognition and Wildfire’s virtual assistant to today’s AI-powered CX platforms.

Tim has worked across IBM, Nuance, LivePerson, Wildfire, and now Sprinklr, giving him a unique longitudinal view of how interface, infrastructure, and customer expectations have evolved. The conversation blends technology history, sales war stories, and practical guidance for executives trying to make sense of AI in 2026.

What Wildfire got right about AI, decades early

Tim describes Wildfire as a groundbreaking virtual assistant that functioned much like a human personal assistant, screening calls, reading messages, and letting users control interruptions through natural voice commands. Users could say things like “Only let my important calls through” or “Play only my messages from Susan Hunt,” capabilities that mirror what we now expect from smartphone ecosystems and smart assistants.

The product was ahead of its time technically and conceptually but constrained by the infrastructure and business model of the era. It tied up expensive phone trunks and depended on carriers who struggled to monetize it, even as the interface and natural language capabilities anticipated what Alexa, Siri, and other assistants would later mainstream.

How voice recognition and NLP paved the way for today’s AI

Tim points to the voice recognition era at Nuance as one of the “wildest rides,” where the industry repeatedly proclaimed “this is the year of voice recognition” until the technology finally reached a level that genuinely changed how people interacted with systems. Natural language processing and understanding became the foundation for today’s agentic AI, enabling people to converse with applications in ways that feel intuitive and human-like.

He notes that what once required painstaking human-built vocabularies—mapping every possible way of saying “yes” or “no”—can now be learned and generalized by AI systems that infer intent from patterns and confidence levels. This shift dramatically lowered the cost and complexity of building robust conversational experiences, even as it raised new design and governance questions.

From cost-cutting IVRs to true customer experience

Both Susan and Tim remember when automation in telcos, banks, and airlines was primarily a cost-cutting play, driven by the inability to scale human agents fast enough, especially during spikes and disruptions. Early IVR systems often frustrated customers so much that public “press 0 to reach an agent” cheat sheets became widespread, signaling that many people were actively bypassing automation to get real help.

Tim credits Nuance and forward-looking carriers with shifting the conversation from pure cost reduction to customer experience outcomes, emphasizing first-call resolution and satisfaction. A pivotal pattern was using automation as triage: collecting trip details for cruise bookings, for example, then handing a fully contextualized call to an agent, which could cut handle time dramatically while preserving the human relationship where it mattered most.

Why empathy and ethics in flows matter

A recurring theme is the importance of ethical and emotionally intelligent design in automated journeys. Tim recounts a workshop on an insurance claims chatbot where someone finally asked, “If this person is standing by the side of the road, shouldn’t we first ask if they’re okay or need help before we take their claim?”

This single question turned into a pattern similar to the “If this is a medical emergency, hang up and dial 911” message, prioritizing safety and humanity before transactional efficiency. At Sprinklr, the idea that “people remember how you made them feel” is treated as a core principle, recognizing that perceived respect and care directly influence long-term customer loyalty.

Another story involves a bank call center agent who literally left the office to give cash to a stranded customer whose ATM card had failed when he urgently needed gas to get to the hospital. That anecdote reshaped how Tim thinks about lost-card flows: beyond automating card replacement, systems should also proactively ask “Do you need money right now?” and offer instant digital disbursement where possible.

Why AI implementation is still so hard for enterprises

Despite decades of evolution, Tim sees AI implementation itself as today’s biggest obstacle for executives. Business leaders are bombarded with mixed messages that portray AI as both an existential threat to jobs and a silver bullet, leaving many hesitant to deploy it in meaningful ways.

Under the surface, large enterprises still run on “dinosaur” systems, especially in billing, policy, and core operations. Plugging AI into these environments is non-trivial: even something that seems simple, like exposing a customer’s balance, requires deep integration into legacy processes and systems. Susan notes that no one she talks to can articulate a universally obvious AI roadmap for a major bank or telco, reinforcing the sense that we are in a true “Wild West” period where roadmaps are customized, experimental, and evolving.

A realistic starting point: AI as Agent Assist

One of the most practical use cases Tim highlights is Agent Assist, where AI listens to conversations or reads chats in real time and surfaces relevant knowledge articles, next-best actions, and contextual guidance to human agents. He compares it to replacing the giant three-ring binder of answers on a rep’s desk with an intelligent, responsive sidekick that understands the conversation and fetches what is needed.

Agent Assist does not replace agents; it reduces cognitive load, search time, and training ramp while preserving human judgment and empathy. For risk-averse organizations, this type of supportive AI can be an effective first step that demonstrates tangible value without handing full control to automation.

What is overhyped about AI right now

Tim sees one of the most overhyped beliefs as the idea that AI will “automatically take over everything” inside a company once deployed in one corner of the business. He recounts an airline that feared keystroke tracking would lead directly to full automation of agent tasks, as if workflows would somehow self-promote into production without human oversight.

In his view, AI can observe patterns and propose flows—for example, repeatedly answering frequent flyer balance questions—but real deployment still requires people to validate, design business rules, and decide what should and should not be automated. Certain intents, like “I want to cancel my subscription,” are almost universally earmarked to go to a human, illustrating that business strategy and customer retention goals strongly shape where AI is allowed to act autonomously.

Voice, channels, and the future of CX work

Despite constant predictions that voice will die and contact centers will disappear, voice remains the largest channel in customer care, even as digital volumes rise. Tim attributes this partly to generational behavior and partly to the fact that some situations still feel better handled via direct conversation.

At the same time, younger generations are extremely comfortable with self-service and digital-first experiences; Susan notes that even her four-year-old granddaughter can handle most modern interfaces without instruction. The future likely looks less like one channel replacing another and more like channel-agnostic orchestration: brands meeting customers via phone, chat, email, apps, and social, with AI helping to unify context and intent across all of them.

On the workforce side, both Susan and Tim argue that AI-driven job displacement will follow historical patterns seen with technologies like the cotton gin, laptops, and mobile phones. Some roles will be eliminated or transformed, but new categories of work in building, maintaining, and governing AI systems will emerge, and the overall effect will be a reconfiguration rather than a simple subtraction.

Career pivots and “stare down the bull” moments

In keeping with the show’s theme, Susan asks Tim for his biggest “stare down the bull” moment, when pressure was high and the market was shifting under his feet. Tim points to the period at Nuance when he was heavily invested in selling voice-activated voicemail, only to see handset makers and network providers embed similar capabilities directly into devices and core services.

Faced with a collapsing product thesis, he had to quickly pivot from application-centric plays like voicemail and dialing to broader care and conversational automation opportunities, particularly in telco customer service. That shift opened new paths for him and his company but required letting go of a strategy he had believed would be his route to success.

How we will look back on this AI era

Looking 15 years ahead, Tim predicts we will laugh at how convinced some people were that AI would fully replace humans in many domains. He expects continued advancement, broader adoption, and better implementation patterns, but not total human obsolescence.

Instead, he anticipates that we will recognize AI as a powerful but bounded tool that took over expensive, repetitive, and rudimentary tasks while amplifying human capabilities in more complex, relational, and judgment-heavy work. Just as voice never died, human roles will likely evolve rather than vanish, and the enduring differentiator will be organizations that combine automation with empathy, ethics, and clear-eyed understanding of their own systems and customers.

This episode offers a grounded, experience-rich view of where AI in customer experience has been, where it is now, and where it is realistically headed.

 

Connect with Susan: 

LinkedIn

Website 

Connect with Tim

LinkedIn