Generative AI is taking the next big leap from machines that rip, mix, and burn words, images, and videos to agents that can create and follow complex multi-step project plans to complete significant projects and automate major portions of jobs. And that raises a big, big question for us old-fashioned meatbag humans: are we just an outdated model now?
After all, Salesforce CEO Marc Benioff just told the world he cut 4,000 customer service roles because AI agents have reduced support time and costs.
“I need less heads,” he told the podcast The Logan Bartlett Show.
Global AI spend will hit $632 billion by 2028, the IDC says, and the IMF thinks AI will affect almost 40% of all jobs globally. So this is a big deal for everyone.
The new AI agents will write blog posts, answer customer support inquiries, research companies, optimize websites, provide data-driven insights, manage clinical trials, and much more. For the companies launching them, these agents aren’t just copilots: they’re becoming semi-autonomous digital teammates embedded in all of the work a company does.
Here are just a few of the dozens of AI agent pitches I’ve received in recent days:
- HubSpot launched 15 new agents plus an agent marketplace
- Adobe announced six agents and an “agent composer” for companies to control them
- Airtop released a conversational interface to build your own agents just by typing
- Unily has launched an “Agent Orchestrator” to manage AI agents
- SK Telecom in Korea just announced a partnership to develop AI agents
- Grove AI has launched an agent for clinical trials
From a certain perspective, these AI agents could be seen as a new kind of robot Terminator, one that kills jobs instead of people. Industry analyst Jeremiah Owyang, for instance, has a theory that we will soon see an “autonomous organization” created by AI or AI agents that doesn’t need any people at all.
“It will offer services that people will purchase, it will generate revenue on its own, it will operate without any humans at the helm,” Owyang suggests.
He’s reading the tea leaves well. That’s likely to be some Silicon Valley techno-nerd’s paradise of the future, and it’s probably being built right now. But it’s not likely to be worker friendly or pro-social. At least, not without some safeguards.
For others, the whole AI agents movement is just the next step in the evolution of work.
“Like the Industrial Revolution didn’t eliminate farming, AI won’t eliminate entire jobs,” Airtop CEO Amir Ashkenazi told me last week, noting that before the industrial revolution, 80% of humans worked in farming, while now it’s less than 2%. “Every revolution has transformed the job market.”
Airtop’s contribution to the revolution is “democratizing automation” via a conversational builder that helps you build your own agents, perhaps to reclaim some of your power as opposed to the ones your company has bought to automate key parts of your job.
Ashkenazi, for example, uses his own platform to automatically fetch transcripts for all his meetings, forward them to ChatGPT Assistant, get back a summary and list of action items, and auto-draft a follow-up email for all participants. All he has to do is check the email, perhaps adjust a few things, and hit Send.
Your role as a human, then, is orchestrator, manager, decider.
“AI is turning everyone into a manager of agents,” he told me after a dinner in San Francisco. “In the future, careers will be defined not by how many tasks you can do yourself, but by how effectively you orchestrate a fleet of AI agents around you.”
HubSpot seems to take a similar tack, although they’re selling AI agents for companies, not specifically people. I chatted with the company’s head of product, Karen Ng, at HubSpot’s product conference recently.
“We believe in hybrid teams,” she told me. “I don’t think you can replace human creativity and taste.”
The company’s CTO Dharmesh Shah, who released Agent.ai, the “#1 Professional Network for AI Agents” last year, says this is the decade of agents. (I accidentally typed in Agents.ai when researching this article, only to find my “new digital assistant,” the interestingly named “B*tch Bot.”)
Shah sees agents as force multipliers: making each human better, stronger, faster, smarter.
“The goal is to build with the machine,” he said last week in San Francisco. “To help take you … to the power of AI.”
His advice, then, is to list all the things you have to do and accomplish, and try them with AI. Not all will work yet, he says, emphasizing the yet. Shah referenced a study (likely this one) that demonstrates that AI improves human performance: elevating an individual with AI over the performance of a team of humans.
But that a team of humans with AI does the best of all:
“They found that an individual using AI outperformed in terms of coming up with better solutions to business problems than a team with no AI,” Shah says. “So the individual does become that super contributor with AI. What they also found, and this is less surprising, that teams with AI produce the best results and most creative results to business problems.”
Which is aligned with the now-familiar saying: AI won’t take your job, but someone using AI will.
The human part is critical, and I can see that from what Salesforce has become, minus the humans, having recently received an email from “Agentforce Sales Agent” which wants to help me help Forbes streamline processes and grow. I didn’t bother explaining that “I just write here,” I just unsubscribed.
(I also repeatedly emailed Salesforce PR about CEO Benioff’s agent and workforce comments, to no response. Which can only mean, I presume, that Salesforce is deploying agents in entirely the wrong ways.)
As HubSpot head of product Ng freely admitted, AI can make huge mistakes. In the LLM age of probabilistic answers, not deterministic responses, AI agents will be both amazing and awful, on occasion, and humans need to be in the loop.
There’s a lot of good available from AI and the developing AI agents. Automating work that is tedious and dull is a good thing, as long as we can repurpose people to more human tasks.
And it can level playing fields:
“One bright spot is how AI lowers the barrier to entrepreneurship,” says Ashkenazi. “People can develop software faster than ever, and with platforms like Airtop, they can launch custom marketing agents to reach their audience. Unicorns with fewer than ten employees will exist very soon—AI makes it possible.”
Of course, as I told him, everyone loves equalization when it means taking them from lower on the food chain to higher. Few love it as much when leveling the playing field means bumping them down a few notches.
The reality, however, is this: AI is here. AI agents are here. They are only getting better, smarter, faster, and more capable.
Perhaps now, as ever, the mandate is simple: adapt or die.
In which case your job as a human is to find the best AI agents and customize them to your needs and goals. And to find ways to make everything you do better as a result.
Brave new world?