Featured
Artificial Intelligence
AI In Customer Experience Needs Trust, Not Just Automation Ft. Anshuman Singh, CEO, EMEA At HGS
Share
Overview
In this episode of TechDogs Discover Dialogues, host Jemish Sataki, Senior Content Lead at TechDogs, sits down with Anshuman Singh, CEO, EMEA at HGS, to discuss how enterprises should rethink customer experience in the AI era.
The conversation explores customer experience strategy, assistive AI, agentic AI, human empathy, customer trust, enterprise AI governance, reversibility, staged rollouts, contact metrics, knowledge management, and the leadership mindset needed to guide large-scale CX transformation.
At the center of the discussion is a practical message: AI should not be used to automate every customer interaction. Businesses must understand which parts of customer experience are routine and transactional, and which moments are emotionally charged, high-stakes, or deeply meaningful.
Anshuman explains that enterprises often organize customer experience around channels such as websites, chat, email, or phone support. However, the more important distinction is not the channel. It is whether the experience is informational, transactional, or deeply meaningful.
Many transformation programs optimize the contact instead of the outcome. This leads to customer journeys where people are kept on hold, sent back to websites, or forced through processes that do not solve the actual problem. The better approach is to focus on what the customer is trying to achieve.
One of Anshuman’s strongest points is that customer experiences are becoming K-shaped. Deeply meaningful experiences are moving upward because they require emotional connection, context, and trust. Transactional experiences are moving downward because they are increasingly being automated.
The problem arises when enterprises automate the wrong part of the experience. If a business automates a deeply meaningful moment, it risks damaging trust. If it automates a routine interaction well, it can improve speed and convenience.
According to Anshuman, human involvement remains essential in emotionally charged or high-stakes situations. A customer may begin with a transactional issue but show signs of distress, frustration, churn risk, or urgency during the conversation. That is where signal-based escalation becomes important.
Many customer service systems still use task-based escalation. If a customer presses a button to cancel an account, they may be sent to a human. But if the customer starts with an order issue and later becomes likely to cancel, the system may not detect that shift. AI-led CX must become better at understanding signals, not just tasks.
The conversation also explores why many AI pilots fail when they move into enterprise environments. It is easy for someone inside a company to buy an off-the-shelf AI tool, create a demo, and show potential. But moving from demo to full enterprise deployment requires crossing several gaps.
These include governance, authoritative knowledge, testing, cost control, orchestration, model selection, and understanding the blast radius of AI.
Anshuman shares an example where an AI system gave three different answers when asked the same question at different times. The initial reaction was to assume hallucination. A deeper review showed that the AI was relying on three different knowledge base articles that contained three different answers.
In that case, the AI was not the core issue. The knowledge corpus was. This highlights an old but critical discipline: knowledge management. Enterprises need clear ownership of authoritative answers. Without that, AI will process inconsistent information and produce inconsistent outcomes.
Enterprise AI is becoming more complex as organizations move from single agents to multi-agent systems. Anshuman uses the example of an IT help desk, where a triage agent, knowledge base, and process agent may need to work together. Just because one agent works well does not mean the overall customer outcome will be achieved.
That is why testing agentic orchestration is critical before scaling AI.
Another practical point from the conversation is that not every problem requires AI.
Some issues can be solved through simple automation, better process design, better knowledge management, or clearer customer journeys. Similarly, not every AI problem requires the most advanced model. Using a sophisticated model for a simple task can create unnecessary cost and complexity.
The right model should fit the right problem.
The conversation then moves into leadership. Anshuman believes businesses must stay focused on customer outcomes, not only products or technologies. Companies exist because there is a paying customer at the end. If they stop obsessing over what the customer wants to achieve, they risk becoming irrelevant.
He also emphasizes that businesses exist in a societal context. Leaders must remember the broader role their organizations play, especially when serving public sector or citizen-facing environments.
In his closing thoughts, Anshuman reflects on the debate around AI and jobs.
He believes the future will likely land somewhere in the messy middle. History shows that major technology shifts rarely eliminate the old world entirely. E-commerce grew, but physical shops still exist. AI will change work, but the outcome will not be purely dystopian or purely optimistic.
His advice to young professionals is to stay open-minded, keep learning, and pursue interests even when their immediate use is not obvious.
Anshuman Singh is the CEO, EMEA at HGS. He brings nearly three decades of experience across technology, design, digital transformation, customer experience, digital commerce, customer self-service, and service innovation.
Trained as an architect from IIT Roorkee and later in product design at IIT Bombay, Anshuman began his career in IT during the dot-com era and helped set up the user experience group at Mindtree. After nearly 25 years at Mindtree, he moved into customer experience transformation at HGS, where he focuses on how generative AI, customer service, human empathy, and business process management can reshape enterprise CX.
The conversation explores customer experience strategy, assistive AI, agentic AI, human empathy, customer trust, enterprise AI governance, reversibility, staged rollouts, contact metrics, knowledge management, and the leadership mindset needed to guide large-scale CX transformation.
At the center of the discussion is a practical message: AI should not be used to automate every customer interaction. Businesses must understand which parts of customer experience are routine and transactional, and which moments are emotionally charged, high-stakes, or deeply meaningful.
Customer Experience Must Focus On Outcomes
Anshuman explains that enterprises often organize customer experience around channels such as websites, chat, email, or phone support. However, the more important distinction is not the channel. It is whether the experience is informational, transactional, or deeply meaningful.
Many transformation programs optimize the contact instead of the outcome. This leads to customer journeys where people are kept on hold, sent back to websites, or forced through processes that do not solve the actual problem. The better approach is to focus on what the customer is trying to achieve.
CX Is Becoming K-Shaped
One of Anshuman’s strongest points is that customer experiences are becoming K-shaped. Deeply meaningful experiences are moving upward because they require emotional connection, context, and trust. Transactional experiences are moving downward because they are increasingly being automated.
The problem arises when enterprises automate the wrong part of the experience. If a business automates a deeply meaningful moment, it risks damaging trust. If it automates a routine interaction well, it can improve speed and convenience.
Where Humans Still Matter
According to Anshuman, human involvement remains essential in emotionally charged or high-stakes situations. A customer may begin with a transactional issue but show signs of distress, frustration, churn risk, or urgency during the conversation. That is where signal-based escalation becomes important.
Many customer service systems still use task-based escalation. If a customer presses a button to cancel an account, they may be sent to a human. But if the customer starts with an order issue and later becomes likely to cancel, the system may not detect that shift. AI-led CX must become better at understanding signals, not just tasks.
Why AI Pilots Fail To Scale
The conversation also explores why many AI pilots fail when they move into enterprise environments. It is easy for someone inside a company to buy an off-the-shelf AI tool, create a demo, and show potential. But moving from demo to full enterprise deployment requires crossing several gaps.
These include governance, authoritative knowledge, testing, cost control, orchestration, model selection, and understanding the blast radius of AI.
Governance Starts With Knowledge
Anshuman shares an example where an AI system gave three different answers when asked the same question at different times. The initial reaction was to assume hallucination. A deeper review showed that the AI was relying on three different knowledge base articles that contained three different answers.
In that case, the AI was not the core issue. The knowledge corpus was. This highlights an old but critical discipline: knowledge management. Enterprises need clear ownership of authoritative answers. Without that, AI will process inconsistent information and produce inconsistent outcomes.
Multi-Agent Orchestration Needs Testing
Enterprise AI is becoming more complex as organizations move from single agents to multi-agent systems. Anshuman uses the example of an IT help desk, where a triage agent, knowledge base, and process agent may need to work together. Just because one agent works well does not mean the overall customer outcome will be achieved.
That is why testing agentic orchestration is critical before scaling AI.
Not Every Problem Needs AI
Another practical point from the conversation is that not every problem requires AI.
Some issues can be solved through simple automation, better process design, better knowledge management, or clearer customer journeys. Similarly, not every AI problem requires the most advanced model. Using a sophisticated model for a simple task can create unnecessary cost and complexity.
The right model should fit the right problem.
Leadership In The AI Era
The conversation then moves into leadership. Anshuman believes businesses must stay focused on customer outcomes, not only products or technologies. Companies exist because there is a paying customer at the end. If they stop obsessing over what the customer wants to achieve, they risk becoming irrelevant.
He also emphasizes that businesses exist in a societal context. Leaders must remember the broader role their organizations play, especially when serving public sector or citizen-facing environments.
The Messy Middle Of AI And Jobs
In his closing thoughts, Anshuman reflects on the debate around AI and jobs.
He believes the future will likely land somewhere in the messy middle. History shows that major technology shifts rarely eliminate the old world entirely. E-commerce grew, but physical shops still exist. AI will change work, but the outcome will not be purely dystopian or purely optimistic.
His advice to young professionals is to stay open-minded, keep learning, and pursue interests even when their immediate use is not obvious.
Key Takeaways
- Customer experience should be designed around outcomes, not only channels.
- Routine and transactional experiences can be automated, but meaningful moments need care.
- The Kano model helps businesses identify must-haves, performance attributes, and delighters.
- Customer experiences are becoming K-shaped.
- Personalization is now table stakes, not a differentiator.
- AI is currently strongest in assistive customer service use cases.
- Agentic AI requires stronger governance because it completes tasks.
- Human empathy remains essential in emotionally charged and high-stakes interactions.
- Signal-based escalation is more important than task-based escalation.
- Many AI pilots fail because governance, testing, costs, and knowledge management are not ready.
- Inconsistent knowledge bases can create inconsistent AI outputs.
- Enterprises must understand the blast radius of AI before production deployment.
- Multi-agent orchestration must be tested end to end.
- AI is not always cheaper than human support.
About The Guest:
Anshuman Singh is the CEO, EMEA at HGS. He brings nearly three decades of experience across technology, design, digital transformation, customer experience, digital commerce, customer self-service, and service innovation.
Trained as an architect from IIT Roorkee and later in product design at IIT Bombay, Anshuman began his career in IT during the dot-com era and helped set up the user experience group at Mindtree. After nearly 25 years at Mindtree, he moved into customer experience transformation at HGS, where he focuses on how generative AI, customer service, human empathy, and business process management can reshape enterprise CX.
Thu, Oct 8, 2026
Enjoyed what you've read so far? Great news - there's more to explore!
Stay up to date with the latest news, a vast collection of tech articles including introductory guides, product reviews, trends and more, thought-provoking interviews, hottest AI blogs and entertaining tech memes.
Plus, get access to branded insights such as informative white papers, intriguing case studies, in-depth reports, enlightening videos and exciting events and webinars from industry-leading global brands.
Dive into TechDogs' treasure trove today and Know Your World of technology!
Disclaimer - Reference to any specific product, software or entity does not constitute an endorsement or recommendation by TechDogs nor should any data or content published be relied upon. The views expressed by TechDogs' members and guests are their own and their appearance on our site does not imply an endorsement of them or any entity they represent. Views and opinions expressed by TechDogs' Authors are those of the Authors and do not necessarily reflect the view of TechDogs or any of its officials. While we aim to provide valuable and helpful information, some content on TechDogs' site may not have been thoroughly reviewed for every detail or aspect. We encourage users to verify any information independently where necessary.
Loading comments...