Demand Agents: Do Customers Want AI?

It's the season for new year predictions. I've published my own marketing operations predictions, and seen plenty of tech predictions from analysts and influencers. One common theme among all of them is that 2025 will be the year of the AI agent. There has been a heavy media focus on AI over the past few years, largely because it's the hot new thing in technology circles. Some early adopters have seen success with it, while others have used it as cover for the job cuts they wanted to make anyway. So far, the hype hasn't translated into widespread corporate roll-outs.

The Unspoken Truth

For all the talk of revolutions and digital transformation, it's unclear when mass adoption of Generative AI technology in the enterprise will take place - or even if it ever will. That shouldn't be a surprise. It takes time to build a business case for new technology. Use case definition and requirements gathering are long-winded processes that always move slower than expected. That's even before the CFO can run the cost/benefit analysis. There was plenty of talk through 2024 about the lack of ROI on generative AI projects. We're only just seeing a response to that.

In recent months, there has been a notable shift in the sales pitch from generative AI firms. It has always been understood that AI would eventually replace human workers, leading to job cuts. Yet, no one was willing to say the quiet part out loud. That has changed. Microsoft and Salesforce have started explicitly talking about the financial savings from replacing human workers with AI agents. We're going to see more of such talk because it's the only way that generative AI makes financial sense for anyone. A lot of independent analysts still haven't noticed the change in narrative - partly because some AI vendors are now pitching to take their jobs too. See Google Deep Research as an example of this.

Customer Support

In the early days of the generative AI boom, there was a brief craze for announcing AI related job cuts. Most famously, BT said they would make 10,000 employees redundant by 2030. This wasn't just PR for the markets. These are jobs that companies have been looking to automate for a long time. It's no coincidence that the initial wave of generative AI chatbots focused on customer support use cases, because that's an area where the business case for replacing humans with automation was already well established. Executives had previously decided to replace their first line support agents, and tech firms were already making the requisite job cuts.

It's questionable whether GenAI can truly replace human support, but for many companies it doesn't need to. It just needs to be better than existing customer support chatbots, which Large Language Models generally are. Generative AI can't be any worse than your typical first line support engineer. And, these are not the only customer-facing roles in which people are open to seeing humans replaced with automation. I've previously discussed the impact of AI on translators and creative agencies on this blog. Like customer support, agency budgets routinely face complaints about low customer satisfaction and high costs. Yet, all the talk about job losses often ignores one other career type that people increasingly want to be automated: B2B sales.

The Self-Service Revolution

The traditional VC playbook calls for hiring more and more sales reps as the business grows. There is assumed to be a fixed ratio between revenue and the number of sales reps on the payroll. Other departments can be cut without impacting revenue, depending on the firm's growth stage and product maturity, but never sales. That assumption is starting to come under question. Surveys indicate that B2B buyers are increasingly suspicious of sales reps, and are actively looking to minimise the number of sales interactions when making purchasing decisions. In marketing circles, this has led to product-led growth and self-service purchasing becoming hot topics. Gartner have been pushing both for several years. Neither requires large numbers of sales reps to execute.

Executives are aware of the bloat and the inefficiency of the traditional corporate sales model. Yet, they do nothing about it, because it is seen as the only way to keep customers happy. That is changing. Already, Google have been cutting sales reps in their ad business. Instead, they're pushing customers to use their self-service Generative AI tools to optimise ads. This is an example which others will follow. Already, large amounts of the typical sales playbook are heavily scripted. Many marketers want to cut out the middle man by turning those scripted elements into self-service web tools.

There are aspects of a sales role that can't yet be automated, such as producing quotes - although developers of AI agents will probably disagree. In recent years, a key task for sales has been to gather information about the prospect, and to customise the corporate sales pitch to the specific customer requirements. Generative AI can do that too. Human sales reps will remain in the loop, but fewer reps are needed if Generative AI is writing all their emails and presenting most of their demos. That's not a bad thing.

The Customer Perspective

Twelve months ago, there was a lot of discussion around human-in-the-loop workflows for generative AI. This was mainly pushed by AI firms, who have moved on to selling standalone agents. That's fortunate because human-in-the-loop never made much sense. Sales is an exception to this rule. Sales reps already spend a lot of time rewriting existing content anyway. It's part of the job. Getting AI to help with that rewriting process saves time and leads to better output. Automated sales emails have become a widespread nuisance. They're also incredibly generic. AI offers a potential solution to that problem. It can add more detailed context than a busy sales rep working on multiple deals. That means better quality sales emails for everyone - assuming people actually want to receive them.

The question of chatbots and sales automation illustrates a neglected facet of the AI vs human debate. There has been plenty of discussion about where businesses might want to replace human employees with AI. There has been far less consideration about what customers want, which is ultimately what actually matters. If AI leads to a decline in customer satisfaction, then businesses will be forced to change their approach. We've seen it before with offshore call centres. Some recent reports claim that a slim majority of consumers prefer dealing with AI chatbots over human support agents. I can believe that.

Similarly, the desire to automate B2B sales originates from a change in customer preferences. People have become used to automation. It's simply quicker and far less hassle. Customers still want the option of dealing with a human, but only if they're adding value to the customer experience. Far too many sales reps are a blocker to getting stuff done. At least the chatbot knows something about the products or services it's supporting. As anyone who has ever spoken to BT customer support will attest - offshore call centres frequently don't.