2026: The Year the Robots Still Won't Take Your Job (Probably)

The breathless predictions were wrong. The dismissive ones were too. Here's my two cents on what's happening, why some of it keeps me up at night, and what businesses can do about it.

Three years of AI hype have given us a strange landscape. On one side, you've got vendors promising that AI agents will replace your entire workforce by Tuesday. On the other, you've got sceptics pointing out that ChatGPT still can't reliably count the letters in "strawberry." Both camps are missing what's actually happening.

Most of the AI conversation happens at the extremes. But here on the ground, where actual businesses are trying to get invoices processed and approvals signed off, the picture looks different. The technology that's already here, working quietly in the background, is more consequential than the technology everyone's arguing about.

That's what I want to talk about. Not the science fiction. The science fact. And why some of it genuinely worries me.

The Productivity Shift Is Real

Let me start with something I've experienced firsthand. Eighteen months ago, building a simple PowerApp took me a few weeks. Today, with the right AI assistance, I can get to a working prototype in a day or two. The code review and refinement still takes time. The deployment and testing still matters. But the raw creation cost has dropped dramatically.

This isn't unique to me. Every knowledge worker I talk to has some version of this story. The accountant whose research time has halved. The marketing manager who drafts campaigns in a fraction of the time. The project manager whose reports practically write themselves.

What's uncomfortable about this is that these efficiency gains don't automatically translate into more jobs or money. They translate into the same work done by fewer people, or more work done by the same people. We've been through versions of this before (the spreadsheet didn't kill accounting, it changed it) but the pace and breadth of this shift feels different.

I don't say this to be alarmist. I say it because pretending it isn't happening won't help anyone prepare for it.

Where I Think We're Actually Headed

Based on what I'm seeing in the market and the technology, here's what I expect 2026 to bring.

AI agents will start doing real work, badly. The autonomous agent hype has been running ahead of the reality for years, but the gap is closing. We'll see more businesses deploying out-of-the-box AI agents that handle genuine multi-step tasks: processing applications, triaging support tickets, managing routine communications. Most of these early deployments will disappoint. The agents will make errors that humans wouldn't. They'll miss context. They'll confidently do the wrong thing. But the businesses that learn from these failures will have a meaningful head start.

The middle layer of knowledge work gets squeezed. Tasks that used to require a junior professional (research, first-draft writing, data compilation, basic analysis) are increasingly automatable. I hope this doesn't mean those jobs disappear overnight, but it does mean the pathway from graduate to expert changes. You can't develop judgement by doing tasks that AI now handles. Businesses will need to think deliberately about how they develop their people.

Enterprise software vendors will extract more while delivering less. I've watched Microsoft and Salesforce bundle AI features into premium SKUs while raising prices across the board. The pattern is clear: monetise the captive customer base while you still can. Power BI Pro went up 40% in 2025. Salesforce raised Tableau on-premises subscriptions by up to 33% to push customers toward cloud. These aren't innovations. They're value extraction from businesses that feel locked in. (ref https://www.convivialtools.org/field-notes/whyyournextbirenewalshouldbeyourlast)

Small businesses get genuine access to automation for the first time. Here's the optimistic flip side. Tools that were enterprise-only are becoming accessible to businesses with 10 or 20 people. The $200/month Microsoft 365 subscription includes Power Automate, Power Apps, and enough AI capability to meaningfully change how a small business operates. Most of them don't know it yet.

The Societal Risks I Can't Stop Thinking About

I'm a technology optimist by inclination. I genuinely believe automation can free people from tedious work and let them focus on things that matter. But I'd be lying if I said the current trajectory doesn't worry me. My opinions below.

The labour market disruption is going to be uneven and ugly. There is little to no governance surrounding the implementation and impacts of AI in business. We'll see pockets of severe displacement alongside pockets of labour shortage. The skills that become valuable and the skills that become worthless won't distribute fairly. The 45-year-old operations manager who spent two decades developing expertise in a specific domain might find that domain suddenly automated. The retraining rhetoric sounds good in policy papers. It's brutal in practice.

The concentration of power accelerates. The companies that control the AI infrastructure (Microsoft, Google, Amazon, OpenAI, Anthropic) are accumulating leverage at a rate that should concern everyone. When the tools your business depends on are controlled by a handful of players, you're not a customer. You're a dependency to be optimised.

The information environment gets worse before it gets better. We're already swimming in AI-generated content of dubious quality. Every low-quality article and fake review and synthetic voice call degrades the signal-to-noise ratio a little more. The tools to detect AI content lag behind the tools to create it. I don't see that changing in 2026.

I don't have solutions to these problems. I'm not sure anyone does. But I think acknowledging them honestly is better than pretending they're not real.

What Businesses Can Actually Do

So where does that leave a business owner trying to navigate this? Here's my honest advice.

Get your fundamentals in order before chasing AI. If your data is scattered across seventeen spreadsheets and your processes live in people's heads, an AI tool isn't going to save you. It's going to fail expensively. Clean data, documented processes, and clear metrics aren't sexy but they're the foundation everything else builds on.

Invest in your people's AI literacy. Not just "prompt engineering" courses. Actual, practical experience using AI tools in their daily work. The employees who understand what AI can and can't do reliably will be dramatically more valuable than those who either fear it or blindly trust it.

Audit what you're already paying for. Most businesses I work with are paying for software they barely use. Your Microsoft 365 subscription almost certainly includes automation capabilities you've never touched. Understand what you've already got and do not commit to any software subscriptions longer than 12 months.

Build relationships, not dependencies. When you're choosing technology partners or software vendors, think about what happens if that relationship sours. Can you get your data out? Can you move to an alternative? The vendors making it hardest to leave are usually the ones you should be most wary of.

Start small and learn. Pick one irritating manual process. Something that takes someone an hour a week and drives the team mad. Automate that. See what works and what doesn't. Build your confidence and capability before you tackle the big stuff.

Have honest conversations about roles. If AI is going to change someone's job significantly, they deserve to know. Pretending nothing's happening until you restructure the team is a betrayal of trust that damages your culture far more than an honest conversation would.

Stronger connections, stronger future

I know this piece has been heavier than my usual fare, but it's been bouncing round my brain more of late and it helps to get it down in writing.

The businesses that navigate this well won't be the ones with the most sophisticated AI. They'll be the ones that remember what AI can't do.

AI can process information, but it can't build trust. It can generate content, but it can't have a genuine conversation. It can automate tasks, but it can't care about your customer's problem or notice when a team member is struggling.

In a world flooding with synthetic interactions and automated responses, genuine human connection becomes rarer and more valuable. The business that actually picks up the phone. The manager who remembers what's happening in their team's lives. The supplier who solves your problem because they want to, not because an algorithm flagged it.

This isn't soft thinking. It's strategic. As AI handles more of the transactional work, the businesses that differentiate themselves will be the ones that invest in the relational work. The relationships with customers that can't be replicated by a chatbot. The workplace culture that makes people want to stay. The reputation built on actually doing what you say you'll do.

The technology is going to keep advancing. The hype is going to keep cycling. But the fundamentals of running a good business haven't changed: treat people well, solve real problems, be honest about what you can and can't do.

2026 is going to be interesting. The businesses that thrive will be the ones that use technology to free up more time for the things that actually matter. Not less.

Over to you

I'll leave you with a few points to ponder till next we meet:

• If AI halved the time your team spends on admin, what would you want them to do instead?

• How would you develop someone's judgement if AI handles all the tasks they used to learn from?

• When was the last time a customer chose you because of how you made them feel, not what you delivered?

• Are you preparing your business to do more with fewer people, or better things with the same people?