Awan Roy and Suntosh Panikkar on what AI still needs to get customer support right
You contact tech support because something isn't working. The last thing you need is an AI assistant suggesting the same fix you've already tried, or giving you an answer that works for someone else's system but not yours. What you really want is a solution, or at least a clear explanation of what went wrong and when it might be fixed.
AI can help support teams get there faster. It can dig up similar complaints, find out whether a software bug is already being fixed, and help engineers give customers more useful updates. But knowing what happened in another case doesn't automatically mean it knows how to solve yours.
In a recent interview with Bhavna Dubey from the Times of India Spotlight Desk, Awan Roy, VP and Head of NiCE India and Philippines, and Suntosh Panikkar, Senior Director, Technical Support, CX, NiCE, discuss what AI can take off engineers' plates, where it still needs human help and how the job of solving customer problems is changing.
Q. AI tools have made rapid inroads into software engineering, but customer support seems to be taking a more cautious route. Why is it harder to automate?
Awan Roy: When AI came along, I think all of us thought customer support would be one of the areas most affected by automation. What we have found instead is that software engineering has gone much further ahead.
In software engineering, you can generate code, test it, make changes and test it again before the product reaches customers. You have room to experiment and correct mistakes. In customer support, the situation is different. You are dealing with a specific customer who has a specific problem, and their context matters a lot. You cannot afford to make mistakes because the customer is watching the process as you try to solve their problem.
There is also a difference in the information available. There is an enormous amount of code on the internet and in repositories such as GitHub, which AI can learn from. Support tickets, on the other hand, often sit in customer-specific databases and are not publicly available.
So, customer context and the risk of affecting customer satisfaction make support a more challenging area to automate.
Q. Where is AI already making a difference to the day-to-day work of support engineers? Can you give us an example?
Suntosh Panikkar: AI is helping engineers with repetitive tasks such as retrieving historical cases, summarising tickets, categorising issues and triaging cases before they are assigned. Instead of manually searching through different repositories, an engineer can get suggestions based on the customer's issue description and the information available.
For example, if a product has a known issue and the engineering team is planning to fix it in the next release, the relevant knowledge-base article or engineering task may already be available. AI can bring that information up so the engineer can tell the customer that the issue has been identified and explain when the fix is expected.
It saves time because the engineer does not have to start the investigation from scratch. AI can also identify similar issues affecting a particular customer's environment and bring them to the engineer's attention.
This gives engineers more time to work on complex problems and engage meaningfully with customers, rather than repeatedly handling the same issues.
Q. Can AI also help support teams understand customers better, rather than just help them resolve tickets?
Awan Roy: Absolutely. One way AI helps is by allowing support engineers to understand the product more deeply. Earlier, if they encountered an issue they could not resolve, they would often have to go to the research and development team. Now, if they can point AI to engineering documentation and code, it can help them understand whether the problem could be related to the configuration or the code itself. This reduces their dependence on engineering teams.
AI can also help people prepare better for customer meetings. For example, you can get information about the major support tickets a customer has open, the status of those tickets, product defects that need to be fixed and the software versions in which those fixes are expected.
You can also understand how the customer is using the product and what suggestions might help them get more out of it.
Earlier, people were often confined to their own expertise because they did not have a deep understanding of everything outside their scope. AI can bring together information from different knowledge bases and give people more useful insights. That can make support teams more effective and help them have better conversations with customers.
Q. Two customers can report what looks like the same problem, but the solution may not be the same. How can AI get better at understanding that difference?
Suntosh Panikkar: Context is the next frontier for AI in support. It needs to understand the customer's environment, architecture, configuration and history of interactions. It also needs to understand the customer's business objectives.
Two customers may report similar symptoms, but their systems could be configured differently. If AI does not have the right context, it may suggest an answer that is relevant in one situation but not appropriate in another.
The knowledge repository is important, too. AI needs reliable information to work with, and we need a continuous feedback loop. When AI recommends an article or a solution, the engineer can tell us whether it helped resolve the issue. That feedback goes back into the system and helps us improve future recommendations.
We also evaluate the quality of the solutions and the knowledge articles being used. This helps us identify where the information is useful and where it needs improvement. The aim is to make AI's recommendations more accurate over time.
Q. Should AI eventually handle the entire support interaction, or are there situations where a human should always step in?
Awan Roy: I think AI will increasingly become the first point of contact. It can understand what kind of help the customer needs, triage the issue and direct it to the appropriate place. Customers do not have to wait for a human agent to become available just to get an initial response.
For repetitive issues where the answer is already known, AI can do a good job. It could tell a customer that a defect has been identified and that a fix is coming in a particular software release.
The key is that AI must know when it does not have the answer. It should not insist on giving a response that is not helpful. It must be able to recognise when an issue is unique or complex and redirect the customer to a human agent.
I also believe organisations should first use AI to assist support engineers behind the scenes and validate how well it performs. Once there is enough confidence in the quality of its responses, they can gradually give it more responsibility in customer-facing interactions. You cannot compromise on quality, because a poor response can leave the customer frustrated.
Q. As AI takes over repetitive work, what will make a support engineer valuable in the future?
Suntosh Panikkar: AI gives engineers an opportunity to develop their capabilities faster. They can work across more products, learn more about the systems they support and spend more time on complex issues.
For example, engineers often have to deal with legacy systems that have evolved over many years. Understanding how these systems work can be difficult. AI can help them make sense of the code and documentation, investigate issues faster and develop a deeper understanding of the product.
But engineers still need the knowledge to assess whether an AI-generated recommendation is correct. Access to information alone is not enough. They need to understand the problem, troubleshoot it and take responsibility for the solution.
AI is most valuable when it helps engineers spend less time searching for information and more time solving customer problems. The real opportunity is to help teams learn faster, deepen their expertise and deliver better outcomes for customers. Organisations that use AI to empower their people and strengthen customer relationships will be best positioned to benefit from it.
Roy and Panikkar see the future of customer support as a combination of AI and human expertise, rather than a choice between the two. AI can take over repetitive work and make information easier to access, while engineers can focus on complex problems, develop deeper knowledge and spend more time understanding customers.
Roy expects the transition to fully automated customer support to be gradual, particularly where AI interacts directly with customers. As he puts it, support jobs could become more interesting as AI handles the mundane tasks, leaving people to take on work that requires greater creativity, ingenuity and troubleshooting.
The broader challenge for customer support teams, therefore, is not how much of the process they can automate, but how effectively they can combine AI's speed with human judgement. For NiCE's leaders, keeping customer outcomes at the centre of that transition will be essential to making AI genuinely useful in support.
Disclaimer: This article has been produced on behalf of NiCE by Times Internet’s Spotlight team.
AI can help support teams get there faster. It can dig up similar complaints, find out whether a software bug is already being fixed, and help engineers give customers more useful updates. But knowing what happened in another case doesn't automatically mean it knows how to solve yours.
In a recent interview with Bhavna Dubey from the Times of India Spotlight Desk, Awan Roy, VP and Head of NiCE India and Philippines, and Suntosh Panikkar, Senior Director, Technical Support, CX, NiCE, discuss what AI can take off engineers' plates, where it still needs human help and how the job of solving customer problems is changing.
Q. AI tools have made rapid inroads into software engineering, but customer support seems to be taking a more cautious route. Why is it harder to automate?
Awan Roy: When AI came along, I think all of us thought customer support would be one of the areas most affected by automation. What we have found instead is that software engineering has gone much further ahead.
In software engineering, you can generate code, test it, make changes and test it again before the product reaches customers. You have room to experiment and correct mistakes. In customer support, the situation is different. You are dealing with a specific customer who has a specific problem, and their context matters a lot. You cannot afford to make mistakes because the customer is watching the process as you try to solve their problem.
There is also a difference in the information available. There is an enormous amount of code on the internet and in repositories such as GitHub, which AI can learn from. Support tickets, on the other hand, often sit in customer-specific databases and are not publicly available.
So, customer context and the risk of affecting customer satisfaction make support a more challenging area to automate.
Q. Where is AI already making a difference to the day-to-day work of support engineers? Can you give us an example?
Suntosh Panikkar: AI is helping engineers with repetitive tasks such as retrieving historical cases, summarising tickets, categorising issues and triaging cases before they are assigned. Instead of manually searching through different repositories, an engineer can get suggestions based on the customer's issue description and the information available.
For example, if a product has a known issue and the engineering team is planning to fix it in the next release, the relevant knowledge-base article or engineering task may already be available. AI can bring that information up so the engineer can tell the customer that the issue has been identified and explain when the fix is expected.
It saves time because the engineer does not have to start the investigation from scratch. AI can also identify similar issues affecting a particular customer's environment and bring them to the engineer's attention.
This gives engineers more time to work on complex problems and engage meaningfully with customers, rather than repeatedly handling the same issues.
Q. Can AI also help support teams understand customers better, rather than just help them resolve tickets?
Awan Roy: Absolutely. One way AI helps is by allowing support engineers to understand the product more deeply. Earlier, if they encountered an issue they could not resolve, they would often have to go to the research and development team. Now, if they can point AI to engineering documentation and code, it can help them understand whether the problem could be related to the configuration or the code itself. This reduces their dependence on engineering teams.
AI can also help people prepare better for customer meetings. For example, you can get information about the major support tickets a customer has open, the status of those tickets, product defects that need to be fixed and the software versions in which those fixes are expected.
You can also understand how the customer is using the product and what suggestions might help them get more out of it.
Earlier, people were often confined to their own expertise because they did not have a deep understanding of everything outside their scope. AI can bring together information from different knowledge bases and give people more useful insights. That can make support teams more effective and help them have better conversations with customers.
Q. Two customers can report what looks like the same problem, but the solution may not be the same. How can AI get better at understanding that difference?
Suntosh Panikkar: Context is the next frontier for AI in support. It needs to understand the customer's environment, architecture, configuration and history of interactions. It also needs to understand the customer's business objectives.
Two customers may report similar symptoms, but their systems could be configured differently. If AI does not have the right context, it may suggest an answer that is relevant in one situation but not appropriate in another.
The knowledge repository is important, too. AI needs reliable information to work with, and we need a continuous feedback loop. When AI recommends an article or a solution, the engineer can tell us whether it helped resolve the issue. That feedback goes back into the system and helps us improve future recommendations.
We also evaluate the quality of the solutions and the knowledge articles being used. This helps us identify where the information is useful and where it needs improvement. The aim is to make AI's recommendations more accurate over time.
Q. Should AI eventually handle the entire support interaction, or are there situations where a human should always step in?
Awan Roy: I think AI will increasingly become the first point of contact. It can understand what kind of help the customer needs, triage the issue and direct it to the appropriate place. Customers do not have to wait for a human agent to become available just to get an initial response.
For repetitive issues where the answer is already known, AI can do a good job. It could tell a customer that a defect has been identified and that a fix is coming in a particular software release.
The key is that AI must know when it does not have the answer. It should not insist on giving a response that is not helpful. It must be able to recognise when an issue is unique or complex and redirect the customer to a human agent.
I also believe organisations should first use AI to assist support engineers behind the scenes and validate how well it performs. Once there is enough confidence in the quality of its responses, they can gradually give it more responsibility in customer-facing interactions. You cannot compromise on quality, because a poor response can leave the customer frustrated.
Q. As AI takes over repetitive work, what will make a support engineer valuable in the future?
Suntosh Panikkar: AI gives engineers an opportunity to develop their capabilities faster. They can work across more products, learn more about the systems they support and spend more time on complex issues.
For example, engineers often have to deal with legacy systems that have evolved over many years. Understanding how these systems work can be difficult. AI can help them make sense of the code and documentation, investigate issues faster and develop a deeper understanding of the product.
But engineers still need the knowledge to assess whether an AI-generated recommendation is correct. Access to information alone is not enough. They need to understand the problem, troubleshoot it and take responsibility for the solution.
AI is most valuable when it helps engineers spend less time searching for information and more time solving customer problems. The real opportunity is to help teams learn faster, deepen their expertise and deliver better outcomes for customers. Organisations that use AI to empower their people and strengthen customer relationships will be best positioned to benefit from it.
Roy and Panikkar see the future of customer support as a combination of AI and human expertise, rather than a choice between the two. AI can take over repetitive work and make information easier to access, while engineers can focus on complex problems, develop deeper knowledge and spend more time understanding customers.
Roy expects the transition to fully automated customer support to be gradual, particularly where AI interacts directly with customers. As he puts it, support jobs could become more interesting as AI handles the mundane tasks, leaving people to take on work that requires greater creativity, ingenuity and troubleshooting.
The broader challenge for customer support teams, therefore, is not how much of the process they can automate, but how effectively they can combine AI's speed with human judgement. For NiCE's leaders, keeping customer outcomes at the centre of that transition will be essential to making AI genuinely useful in support.
Disclaimer: This article has been produced on behalf of NiCE by Times Internet’s Spotlight team.
Next Story