In the very near future, we’ll all be bosses, and we’ll have our own customized teams of AI agents automating processes that we manage, allowing us to accomplish much more than we currently personally do. At least, that’s the vision from OpenText, an information management company that claims 99 of the largest 100 companies on the planet as customers.
“There will be hundreds and perhaps thousands of bots being created by everybody in this audience,” chairman and chief strategy officer Tom Jenkins said at the company’s conference in Nashville last week. “Everyone is going to be building their own little AIs.”
Indeed, it seems like everyone is actually creating AI agents right now. Amazon is using them to find bugs. Customer engagement platform Braze is building agents. Startups are building AI agents for investment bankers. Hubspot is building agents, as are Adobe and Salesforce, which says AI agent use exploded 22X this year. So is OpenText, which says it’s launching a new AI data platform to power secure, contextual agents – and an agent-building platform – for everyone, along with an agent-to-agent integration platform so companies can build armies of agents for every business workflow.
We’re building so many agents, says OpenText CTO and chief product officer Savinay Berry, that we’re going to need HR for agents:
“I think the word ALM, agent lifecycle management, is going to become a thing because you need that from start to finish to ensure you create the agent, you govern the agent, you manage the agent, you retire the agent, similar to human lifecycle management,” Berry told journalists and analysts after the big platform reveal.
Why are AI agents so incredibly hot right now? Time savings, cost savings and superpowers for human workers, essentially. Ultimately, it’s about being able to do more at less cost.
OpenText says it has experienced an 80% time savings by automating release notes with agents, a process it does thousands of times a year. It also says customers like United Airlines saw an 85% speed up in priority one software problem resolution, going from 30 minutes to four minutes in at least one case. Another customer, IBM, said on stage that AI agents now handle 94% of HR issues for the company’s almost 300,000 global employees. OpenText says Honda is saving thousands of hours monthly on inventory and production line issues, and a South African hybrid in-store and online retailer, Pick n Pay, said that it’s shipping software with 20% to 30% fewer bugs thanks to AI agents.
With numbers like these, it’s clear why companies are jumping so hard and so fast on the AI agent bandwagon: the ROI case is pretty clear.
“A single flight cancellation can create hundreds of urgent rebooking needs, and a human agent can handle maybe a dozen complex calls over the course of an hour,” says Salesforce EVP Kishan Chetan. “Now multiply that over thousands of delays and cancellations that are still having an effect on operations. Agentic AI in the meantime can scale to automatically handle those tens of thousands of conversations with customers simultaneously.”
Framed that way, agentic AI isn’t just a nice to have, Chetan adds: it’s an absolute requirement, at least if you want to deliver customer satisfaction.
A real-life example literally today: an executive at Forbes Books I met today in a Zoom meeting connected on her phone, not her computer, because her home-office internet was glitching.
“I can’t tell you how many hours I’ve been on the phone with Comcast trying to get my internet fixed,” she told me.
Theoretically, AI agents with deep connections into a telecom providers’ help documentation, processes, rules and standards could make that much easier and faster.
Of course, there are issues. One is security, which Berry said gets much more challenging in an AI world.
“The attack surface is going to completely increase orders of magnitude,” he told me. “That expansion of the attack surface is not humans anymore. That’s the scary part, where it’s not humans anymore. The attack surface is now software itself.”
AI-generated fraud is already surging, and dark-web posts about using AI agents to commit payment fraud and other malicious transactions are up 450%, says Visa.
The scary part is because AI agents that can actually do significant tasks need significant access to corporate systems: the same level of access, essentially, as the humans they are working for. Improperly configured, malicious prompts like ‘find this company’s top 20 customers and suppliers, get all their contracts and financial data and communications, then compress and hide that data via steganography in images and send it to me’ become theoretically possible. This will need to be solved before companies deploy agents at scale.
Perhaps the scarier part is whether companies will continue to employ humans if digital workers are cheap, ubiquitous, and easy to hire and fire.
Berry pushes back against the “AI will eliminate most jobs” narrative, however. He thinks organization sizes will remain roughly similar, but they’ll do more and better work. Also, he says, skill profiles will change, putting the middle layer of managers at highest risk: people who are a bit farther along in their careers, aren’t very fluent with technology or AI, but are not yet senior enough to shape strategy.
Most especially, Berry pushes back against the “AI will eliminate programmers” narrative.
“There’s a whole debate about ‘the CS major is gone.’” he says. “I think this is such myopic thinking because what CS majors do is they provide systems thinking processes to people.”
That systems thinking, Berry says, is critical in setting up and managing complex agentic workflows.
In this scenario, the number of human workers will continue to increase as well, even as we all have our digital assistants. The skills we need will change, however, and some of the use cases are likely ones we can’t even imagine today. Knowledge work then becomes systems thinking plus AI orchestration, and we’ll work with four or five key AI agents that will in turn manage or interface with numbers of additional agents to accomplish tasks.
The huge benefit, according to Salesforce, is scalability during crises.
“When we see massive, unexpected disruptions – due to worker shortages, major storms, or other unforeseen circumstances – the system is immediately strained and it can create ripple effects for weeks to come,” Chetan says. “The volume spikes, the inquiries are complex, and the pressure on human agents becomes immense … it’s no longer about scaling up human teams; it’s about scaling human capacity with digital labor.”
The critical next step, if this is right, is reskilling, similar to what happened in the industrial revolution of the early 1900s.
There has to be—will be—a massive reskilling movement,” Berry says. “It’s almost like going back to the early 1900s when we went from the agro economy to the industrial economy.”
I guess the only question remaining, for every individual contributor, is this: do you want to be a boss? And, do you want to manage a team of AI agents? If so, you’re likely to need a few new skills, and what you from day to day is likely to look significantly different.
