AI Agents and the CEOs

There is something going on in the corporate thought leadership of tech firms ranging from the hyperscalers to SMBs around agents and I’ve got a theory why. In the past few years we have blasted past several AI buzzwords: LLMs, chatbots, assistants/ copilots, RAG. But lately, tech CEOs won’t stop talking about agents. Here’s a sample:

From Matt Garman, CEO of AWS:

“The next step is automating tasks — that’s what agents are all about,” he explained. “We’re at the point where you can have thousands of agents executing tasks, but that gets complicated fast.”

From Thomas Dohmke, CEO of GitHub:

We will see a future where code is co-written by humans and AI agents. The human sets the overarching goal, determines constraints, ensures ethical considerations, and divides the work into small chunks that can be handled by the state-of-the-art model, while the AI agent takes on the grunt work of writing, testing, and refining large swaths of code.

From Mathias Biilmann, CEO of Netlify:

As we enter into an era where agents will interact with our products autonomously, and build with our platforms while consuming our content and experiences on the web and beyond, we need to start considering how to craft our product experience specifically for AI Agents.

We need to start focusing on AX or “agent experience” — the holistic experience AI agents will have as the user of a product or platform.

From Amit Wallia, CEO of informatica:

For example, a human resources agent must know the skills and experiences that are essential for job success, while a product agent must be versed in everything from installation to maintenance. Businesses will need to develop expertise in training and refining agent behaviors with this kind of domain-specific data.

Marc Benioff, CEO of Salesforce, has even weaponized this trend to take on the competition, namely Microsoft, tweeting that“Microsoft rebranding Copilot as ‘agents’” is evidence of “panic mode” while skewering Copilot as “Clippy 2.0.”

Microsoft rebranding Copilot as ‘agents’? That’s panic mode. Let’s be real—Copilot’s a flop because Microsoft lacks the data, metadata, and enterprise security models to create real corporate intelligence. That is why Copilot is inaccurate, spills corporate data, and forces… pic.twitter.com/eZpVwgoSH9

— Marc Benioff (@Benioff) October 21, 2024

Benioff’s tweet not only touts Salesforce’s “Agentforce,” it also underscores what every CEO seems to know:  it’s all about the agents.

Is “agent” (or “agentic”) just another overhyped idea—albeit one with persistent definitional ambiguity? Will agents soon cede place to the next new-shiny? Who knows! But I argue the reason so many CEOs have gone out of their way to talk up agents in the past few months has to do with perceptions of AI’s return on investment (ROI).

Companies are dedicating significant spend to AI—approximately 5% of the revenue of large enterprises (revenues over $500 million) according to one survey by Boston Consulting Group, and yet only 25% claim they are seeing value from their AI investment. Return on investment (ROI) needs to be there for companies to continue investing in AI projects, and squishy productivity promises just aren’t cutting it.

At the risk of saying the quiet part out loud, the way CEOs are talking about agents sure sounds like how they talk about employees—only cheaper! Companies that aren’t investing in AI agents not only risk being left behind technologically as these tools become more powerful and capable; they are also leaving money on the table through missed opportunities for headcount reduction. Sure, this messaging push around agents could have to do with advancements in LLMs, multimodal AI, and reinforcement learning. It might also be the result of competitive pressure and keeping up with the Joneses through thought leadership. But when I hear tech CEOs tout the amazing capabilities of the agents they support, what I hear is: “your AI ROI is finally here, and it is headcount reduction.”

While many AI vendors maintain (at least in the short-term) that AI is intended to be augmentative to people, some executives have been explicit about AI’s human-replacing promise. Sebastian Siemiatkowski, CEO of Klarna, for instance is outspoken with his opinion “that AI can already do all of the jobs that we as humans do. It’s just a question of how we apply it and use it.” Several news stories seem to support Siemiatkowski. The Wall Street Journal, for example, recently claimed that AI is directly responsible for the 5.7% rise in IT unemployment, quoting M. Victor Janulaitis, CEO at Janco Associates:

As [companies] start looking at AI, they’re also looking at reducing the number of programmers, systems designers, hoping that AI is going to be able to provide them some value and have a good rate of return.

Both the WSJ’s headline and Janulaitis’s quote are oversimplifications. Rather than AI, what is more likely is that these unemployment rates reflect ongoing spending cuts implemented years ago in response to the post-ZIRP economic downturn. Important as AI may be in the future, accusing it of engineering department layoffs and hiring freezes is the tail wagging the dog.

In the near-term it is doubtful that agents can meaningfully deliver on so many tech executives’ (usually implicit) suggestions of operational cost savings and cost avoidance. Regardless of whether CEOs are candid or evasive about AI’s potential to reduce their customer’s workforce, the reality of this promise has not yet panned out. For now, AI represents a human assistant and AI agents specifically still require human oversight.

Disclaimer: AWS, Microsoft, GitHub and Salesforce are all RedMonk clients.