In the recent episode of the All-In podcast, the 'besties' in the podcast elaborately discussed how AI will destroy certain sectors and tasks. "I think it's now becoming really clear that call centres are going to be the first really big disruption caused by AI," said entrepreneur and investor David Sacks. "I mean all the level one customer support is going to get replaced by AI."
With options to evade human interaction, people are increasingly looking to find ways to solve problems through a machine or AI interaction.
Investor Jason Calacanis, added that people would rather go on YouTube or ChatGPT and ask how to fix things. "I don't want to waste another person's time, just give me the answer as quickly as possible," he said.
If you look at the travel industry, for instance, a sector that heavily relied on human agents at one time is almost completely replaced by AI chatbots and agents powered by generative AI. Thomas Cook India recently announced a new AI-powered platform.
Interestingly, last year, AIM got in touch with Tapan Barman, co-founder and chief executive at Mihup.ai, a conversational AI company, who said that generative AI automation in the BPO sector can not only help streamline repetitive tasks, but enhance decision-making and improve customer services.
In the same podcast, entrepreneur and investor Chamath Palihapitiya spoke about sectors that were believed to remain 'untouched'. For example, with industry giants like Salesforce dominating enterprise software, the idea of AI replacing its core functions once seemed unimaginable.
However, the recent Klarna incident, where the fintech company replaced 50% of their workforce and ended ties with SasS players Salesforce and Workday, proved even legacy sectors as indestructible.
While enterprises such as Oracle, Salesforce and others may have implemented the next big thing, which is AI agents, the adoption is largely vertically integrated, and limited to semi-autonomous capabilities.
"This idea of sort of building agents that can be grounded in content is a great demo but actually not necessarily an impactful product. That's the air of technology we've tried to solve," said Brett Taylor, the CEO of Sierra and board member of OpenAI, in a recent podcast.
Taylor speaks about existing models such as RAG that are meant to improve LLMs, but are often inefficient when it comes to delivering meaningful customer experiences in practice.
He cites an example of how customer interactions with businesses involve taking action rather than answering questions such as subscription upgrades, returns, etc, and the process involves multiple steps and interaction between different systems, which is not possible in the current scheme.
Despite the limitations, companies have been implementing AI across platforms. Big-tech giants such as Amazon, which have implemented AI solutions and products across verticals, have not left their shoppers behind. The company launched their gen-AI powered chatbot 'Rufus' a few days ahead of their upcoming sale.
Conversation with Rufus AI chatbot on Amazon app
Similarly, platforms such as Swiggy and Zomato have also integrated their versions of AI models.
Many companies are implementing AI in one form or another, which seems to be a silent revolution, with many not realising it. Lately, sectors such as HR and legal have also seen major AI implementations. A Bengaluru startup, Spire.AI, recently developed an AI-powered assistant to help enterprises manage hiring and talent development.
Speaking about the advancements in foundation models with voice integration, Sacks believes that call centres can easily be replaced with AI capabilities. He also said that within the next two to three years, there would be a massive disruption in the field.
Not just the capabilities, the training of foundation models, for this sector becomes an added advantage. "One of the reasons why I'm bullish on this customer support use case is because there's a very large data set to train on," said Sacks.
Sachs believes that the product documentation created by companies, along with previous email support and recorded calls, provides a substantial body of data to train the AI model. Also considering how this data is not highly proprietary; it does not include sensitive information such as medical records or confidential legal documents, which makes the data readily accessible.
