I just completed nearly 60,000 miles of travel across Europe, Asia, and the Middle East meeting with hundred of companies to discuss their AI strategies. While every company’s maturity is different, one thing is clear: AI as a business tool has arrived: it’s real and the use-cases are growing.

A new survey by Wharton shows that 46% of business leaders use Gen AI daily and 80% use it weekly. And among these users, 72% are measuring ROI and 74% report a positive return. HR, by the way, is the #3 department in use cases, only slightly behind IT and Finance.

Budgets, as you may expect, are also going up: 23% of large companies are spending $20M or more, and 43% are spending over $10M per year.

What are companies getting out of all this? Productivity. The #1 use case, by far, is what we call “stage 1” usage – individual productivity. AI is helping people summarize meetings, analyze data, find information, and author or analyze documents. These personal productivity use cases are very real, but they’re also only the beginning.

Gen AI May Be The New Microsoft Office

I hate to say it, but these are similar use cases to the early days of word processing, spreadsheets, and internet search: they’re “individual productivity” benefits. Microsoft really designed for this. The MS Copilot is slowly becoming the “New Microsoft Office Suite.”

Of course AI does much more. About 12% of companies now have corporate agents (ie. IBM’s “Ask HR” for example), and these “knowledge and information management” chatbots will be everywhere. Internal AI agents like this can replace complex portals and Sharepoint sites, and also serve as customer support systems. Every company will have one.

One of our clients, a large healthcare company, has been running an employee chatbot (agent) for a four years. It’s so successful that every HR application is now being integrated behind it. Employees ask it for help with pay, benefits, work schedules, and even training.

Recruiting is another proven use of AI: job candidates can chat with an agent, take an AI-based assessment, and get interviewed by an AI avatar. And they can do this in the middle of the night: no need to schedule a call with a recruiter or hiring manager.

While the high-ROI multi-function agents aren’t quite here yet (stage 3 above), companies are now deploying AI-based coaches and AI based learning tools. Many of our large clients are now implementing AI-native learning systems and seeing 30-40% reduction in staff with vast improvements in workforce enablement.

For HR overall our Galileo business has suddenly exploded as more and more companies want a highly tuned HR and learning agent for staff and managers. (Galileo is now a digital HR business partner and Supertutor.)

Crossing The Rubicon: What Chasm Did We Cross?

Crossing a Rubicon means “passing a point of no return.” Well that’s where we are.

Despite inflammatory stories about AI ruining our lives and careers, Gen AI is a useful, pragmatic, easy to understand tool. It’s by no means perfect (listen to my podcast on the high rate of errors from ChatGPT), but once you learn how to use it (and build a trusted data set to train it), it works quite well.

Two years ago the NYT was trying to frighten us with stories of AI acting as a romance partner. Well those stories are over, and thanks to a $Trillion (literally) of capital investment in infrastructure, engineering, and power plants, this stuff is reasonably safe.

I’m not saying it’s 100% safe: if you aren’t careful it’s easy to come up with a wrong answer, a poorly written report, or bogus finding. But we’re getting used to “checking” the AI, so we’re more comfortable with its “probabilistic nature.”

And there are new issues: we’re all going to be living next to a powerplant or a data center, and that raises new political issues.(One of the leaders in the UAE informed me that every ChatGPT query consumes four liters of water, which is now the next big challenge to address.)

Where Do We Go Next?

As I will be discussing in a major webinar this week (AI Trends for 2026), this is barely even the beginning. We have many changes yet to come, so let me mention them here.

From Single User to Multi-Function Use Cases

First, the big ROI from AI is going to be what I call a “multi-functional agent.” (see stage 3 above)

The productivity tools we have today are like having an AI tool to help you turn the steering wheel in your car. (Power steering.) It’s a nice thing to have, but we really want the AI to “take me from place to place,” not “help me steer.”

This is starting to happen in recruiting and training, where we now have agents that can write job requisitions, chat with candidates, schedule interviews, and screen resumes all in one. Soon they’ll be connected to onboarding and annual performance reviews. This “hiring and career” agent is what we call a multi-functional AI agent, and we’re working on a blueprint to help vendors and buyers build these out.

Companies don’t want hundreds of agents running around pulling on steering wheels: they want “smart” agents working on end-to-end business processes. (ie. “design to build to distribute to sell”, or “position to target to market to close a sale” and then “bill to collect to renew and support.” So the individual use-cases we’re exploring will soon be much more integrated.

And as these multi-functional agents appear (many will be built by IT, not by vendors), we then have to change every job in the company. No more need for “interview schedulers” or “SDR appointment creators” or “accounts receivable experts.” The agents do these “steering wheel” things in the context of a workflow.

We’ve seen this is in our Galileo business. What started as an HR “assistant” can now answer a question, develop a course, and give a user the answer to a complex pay, rewards, or internal company question. Galileo literally “builds solutions” for you, taking you from “idea or problem” to “proven solution” like the self-driving car.

Agents Will Have Memory and Personalities

The second change coming is Agents who “know you.” Galileo now remembers who you are and what you did last. Rather than starting from scratch each time, the agents “learn from your usage” or “learn from the business itself.” This makes them more autonomous, personalized, and valuable.

Suppose you’re a manager and you have a capacity problem. You ask Galileo something like “can you help me hire a new staff member?” and Galileo may say “before I open a requisition can I ask you to explain what you want this person to do?” It may then ask you about span of control (it has benchmarks) and say “for the salary you need it may be better to look for internal candidate, do you want me to find someone in the company with the skills you need?”

A month later when you ask Galileo for help again it’s likely to say “it looks like your last hire didn’t ramp as fast as we expected, should we look at a new development plan for your team before we add a new person?”

See where I’m going?

Once all the “steering wheel” agents are working it’s time to put them together and let the AI help steer the entire operation. This is not far away at all, we’ll see a lot of this in 2026.

Data Management Will Now Be Mission Critical

When we run across companies with experience we always find the same thing: the biggest new discipline they developed was data management, data labeling, and data governance. We learned this lesson with Galileo. If the data is not correct, current, and well labeled, the AI will not perform accurately.

Remember the AI has no idea what these words or numbers mean. It simply uses probability and vector calculus to “produce answers.” Even the tiniest error in your data set could cause a high percentage of mistakes (Read my article about 45% of news queries being erroneous for more). So over time companies like IBM, Walmart, and BMS find that data ownership has become mission-critical.

IBM has more than 6,000 HR policies in its “Ask HR” agent, and every policy has an owner who is responsible for keeping it up to date. Now IBM is building an agent to ripple through the policies and monitor regulatory changes in thousands of jurisdictions to point out possible problems. We’ll all be going down this learning curve.

Agents Will Talk to Agents

And there’s more. Now that we’ve “crossed the chasm” we need agent-to-agent communications. These protocols (a2a and MCP) are not mature yet, but companies are working on it. We now have Galileo connected to SAP’s Joule, for example, and more integrations are coming.

Let me warn you, by the way. Don’t get excited and buy 50 different agents for use-cases in HR. The utility of the individual agents won’t seem so exciting if they don’t work together. Many of our clients now sign one-year contracts just to make sure they’re not stuck with something that gets obsoleted in a hurry.

Early next year we’re launching an Agentic AI Blueprint for HR, to show you where these agents should connect to each other. This kind of blueprint will help you make sure you don’t wind up with too many vendors, each pulling the car in different directions. (The self-driving care analogy works pretty well.)

Vendor Risks To Consider

There are still risks ahead. We don’t know if OpenAI is going to clean up its act or not. The MS Copilot is spread thin and going in many directions. And tools like Gemini and Anthropic are going to have to deal with Grok, Deepseek, and others. And if the stock market crashes a lot of these companies are likely to be consolidated.

I think products that focus on quality, pragmatic business areas like Galileo, Paradox, Eightfold, Sana, Arist, and others are ready for purchase. So there are many strong HR vendors in the market.

The HCM vendors, from SAP to Workday to ADP and HiBob to ServiceNow, are building agents into their payroll and workflow engines. They aspire to be your end-to-end multi-functional agent provider too (SAP’s acquisition of SmartRecruiters, Workday’s acquisition of HiredScore, Paradox, and Sana) so we have to keep up with their offerings.

What About Other Fears? Job Loss? Dumbing Down Workers?

Throughout these meetings I heard all sorts of fears. HR staff are worried about their jobs. Recruiters aren’t sure which candidates are real. Some asked me “are we just going to become dummer?”

Here’s my answer to these issues.

If you don’t lean into this revolution, it’s just going to start without you. This is a revolutionary time in business, and we have a once in a lifetime to re-engineer what we do. Now is not the time to chicken out. Get your hands on the tool of your choice or let us show you how to use Galileo. Your fluency and experience will show you new opportunities for your career.

As far as AI taking over our human jobs, I’d stop listening to the techies. There’s just no way.

Even if we do wind up with self-driving cars, we’ll then look back and say “driving wasn’t that great anyway” (I will still have my sports car.) Let’s spend some time focused on other parts of our lives, adding value in new ways.

And AI is so new, so imperfect, and so rapidly changing that we all have jobs as Superworkers, consultants, and innovators to find new use-cases.

When the first spreadsheet was launched in 1981 I remember people thinking that accountants would go out of business. Ha! Look what happened! There are more accountants than ever, they’re just not wasting time adding numbers in tiny columns by hand.

And those of you who are designers, creators, authors, or analysts – think of AI as your own personal supercomputer. Just like cabinet makers use routers and automatic saws, you can still build gorgeous and elegant things, just learn to use the new tools.

Welcome to the new world: the mist has lifted and AI is here to stay. Let’s all get on the Superworker road and help our organizations learn, apply, and leverage this amazing new technology. It’s now all up to us.

Additional Information

AI and HR Trends for 2026: Webinar

BBC Finds That 45% of AI Queries Produce Erroneous Answers

AI: Not Always Right, But Seldom In Doubt (podcast)

Galileo: The World’s Trusted Agent for Everything HR