Most people don’t realize how much better AI got in the last year

Not in the “AI writes nicer emails” way.

In the “AI can now actually do real pieces of someone’s job on its own” way.

That’s the kind of progress that doesn’t go viral on social media.

But it absolutely shows up in company budgets, hiring plans, and decisions about which software tools are still worth paying for.

The next wave of AI disruption won’t feel like a flashy new feature. It will feel like entire job tasks getting cheaper, faster, and more automated than most software companies were built to handle.

The year software work changed

One clear way to see this shift is through software engineering tests.

These benchmarks measure whether AI can fix real coding problems, not just talk about them.

In early 2024, the best AI systems could solve only about 20 percent of real-world coding tasks on a benchmark called SWE-bench.

Roughly a year later, that number jumped to around 65 to 74 percent.

Even if you do not care about coding, the trend matters. In just 12 months, AI went from sometimes fixing a bug to often completing an entire task on its own.

That changes what automation means inside companies.

Instead of simply helping a human work faster, AI can now sometimes complete the full assignment. And software engineering is just one example. Recently we also wrote about the 5 jobs AI will replace in 2026, due to AI disruption.

AI agents are digital workers

This is where the idea of AI agents becomes important.

Especially the recent AI tool from Anthropic.

An AI agent is not just something that answers a question.

It can take a goal, break it into steps, use different tools, and continue working until the task is finished.

For example, instead of drafting a response to a customer, an agent could solve the issue, update company records, notify the customer, and log everything automatically.

That is not a writing assistant, but a digital worker.

Large software companies are already building toward this model. Instead of selling tools that help an employee, they are moving toward systems that do the job themselves.

That represents a major shift.

Traditional software makes a human more productive. Agent-based software can reduce how many humans are needed in the first place.

Why this matters for software companies

For years, software companies made money by charging per employee.

The more workers using the tool, the more revenue they generated.

But if one employee using AI agents can now do the work of three or four people, companies will not need as many software seats. SaaS disruption is inevitable.

That creates tension.

Software companies do not just need to add AI features. They may need to rethink how they charge customers entirely.

Instead of pricing per user, they may need to price per completed task or per outcome. That is a very different business model.

Where disruption will likely hit first

The first areas to change will probably not be flashy creative jobs.

They will be repetitive, process-heavy roles in what companies call the middle office.

These include customer support ticket handling, procurement workflows, finance operations, sales operations, compliance checks, and contract review.

These jobs often follow clear procedures. They involve moving information between systems and checking boxes.

That is exactly the kind of work AI agents are becoming good at.

Right now, AI spending is still a small share of total software spending. But it is growing quickly.

That combination, low penetration but fast momentum, is often how major shifts begin.

Investors are already reacting

Markets do not wait for perfect proof.

They move when a story becomes believable.

In early 2026, stocks tied to office-based coordination work, such as commercial real estate services, fell sharply on concerns that AI could reduce the need for labor-heavy operations.

Logistics companies experienced similar pressure after claims that AI freight tools could automate parts of their workflow.

What makes this more interesting is that many of those same companies are also racing to build their own AI agents.

The concern is simple. If an industry makes most of its money coordinating people and processes, and AI can automate coordination, then profit margins are at risk.

The bigger shift

It helps to think of AI agents as the new interface for work.

Instead of humans clicking through dashboards and moving information around, AI can increasingly handle those steps itself.

If you are buying software, the important question becomes whether the tool completes tasks from start to finish or simply assists a human along the way.

If you are selling software, the more uncomfortable question is whether your business still works if customers need fewer humans using your product.

This is bigger than a race to add features.

It is a shift in how work gets done.

That is why we believe AI is one of the 7 disruptive technologies that will change the world. Shaping the next decade and further. Recent AI disruption in many markets is just the beginning.