Agentic AI will be more impactful than Generative AI for customer service - if it can survive the hype
“You can feel it in the air,” a tech industry contact whispered to me last week, “Generative AI is old news. Everyone is talking about Agentic now.”
Google Trends seems to agree. The below chart shows keyword search trends for “Generative AI” (blue) and “Agentic AI” (red).
Interestingly, the initial blip of interest for both terms started in the last week of September (in 2022 and 2024, respectively). By the first week of December 2024, there were around 50% more searches for “Agentic AI” than there were for “Generative AI” at the beginning of December 2022. The hype train is back, and this time it has a turbo-boosted engine.
From the perspective of a customer service leaders, arguably some of this hype is worth it. By being able to complete whole tasks to achieve higher level goals, an AI Agent is better suited to handle many customer service requests than Generative AI alone.
For example, by carrying out a range of tasks, an AI Agent could be asked to:
Help a customer diagnose a fault with their internet connection and put in place the most efficient solution to fix it
Rebook a customer onto a replacement flight where the aircraft type and seating arrangement have changed
Change a customer’s address details across multiple different systems of record that have not been integrated, each with slightly different fields in their data structure
The demos from Anthropic and others look stunning, and clearly demonstrate the capabilities of a bot that is not just using language, but also logical reasoning, arithmetic, and actions in order to solve problems and complete tasks.
But before going all-in on Agentic AI as the answer to all problems, it’s worth taking a step back to think how use of Generative AI has developed over the past two years. In BCG’s recent report, The Leader’s Guide to Transforming with AI, our research identifies that only about a quarter of companies have been able to realise meaningful value from AI. For many companies, it will feel like they are only just getting to grips with Generative AI, and are now feeling compelled to pivot to Agentic.
So, based on my experience of the last two years, here are some thoughts on how to cut through the hype and focus on activities that will create the most value.
1. Pick use cases that don’t require perfection
The biggest challenge with Generative AI is that, whilst it is relatively easy to stand up some pilots, it is notoriously difficult to scale effectively. Putting the right tech infrastructure, data, ways of working and controls in place to make a bot work as effectively in production as it did in the demo is where the bulk of the effort needs to go. This will be even more the case for AI agent, which we expect, eventually, to be able to complete complex tasks without supervision.
A good way to start is to pick a small number of use cases with a wider acceptable margin of error for accuracy. For example, processes where a human will check the results downstream. This enables implementation and learning to happen without having to wait to achieve perfection.
2. There may be quicker and cheaper solutions to achieve the same outcome
Watching demos of new technology is often inspiring. An interesting and fruitful technique is to use that inspiration to think of new ways to use existing tools. For example, watching an AI agent carry out a process may reveal that there are redundant steps in that process which can be removed.
3. A lot of things will be called “Agentic AI” - some of them are not
Already, many tech companies are rebadging existing tools as “Agentic” or as “AI Agents.” It’s an understandable move, based on the search engine data above. The same rule applies here as it always has, since the early days of automation. Start from the problem that needs to be solve, not from the name of the tool to be used.
4. It is usually better to launch something that works than constantly be chasing the next best thing
I’ve spoken to many customer service leaders in the past year who feel a sense of paralysis caused by the speed of innovation. “What is the point of investing in a technical solution now,” so the logic goes, “if it will be superseded by something better in a few months time.”
A useful insight from the BCG report linked above is that the most successful companies consistently focus on a small number of use cases, and on finishing the job before moving on to something new. They worry less about what is coming round the corner, because they know they will get measurable value from what they are doing today.
It’s going to be fascinating to see what happens with AI agents throughout 2025. One final recommendation is to dedicate time each week to reading and learning about how they work and how their capabilities are developing.
Recommended news articles
CNET: AT&T's new guarantee promises to compensate users if there are outages
Jing Daily: Luxury’s customer service crisis: why brands are failing in China
Forbes: Understanding and preparing for the 7 levels of AI agents
Latest perspectives from BCG
How AI agents are opening the golden era of customer experience
The convergence of AI-powered agents and innovative hardware will enable brands to deliver superior customer experience at a much lower cost-to-serve.
The agents will make customer interactions more personal and less tedious, while the evolving hardware will blend seamlessly into their day-to-day lives.
This article outlines a set of priorities for CEOs to help their companies win at this new frontier