What I am watching in AI: 2026

👋 What’s up friends!

This is Prompt Punk — the best AI sales newsletter 😉

This is my first newsletter of the year. It’s a short one focused on what I am looking forward to in AI in 2026. Here goes…

2026 is here, and my relationship with the AI news cycle has changed.

I am less perturbed by daily headlines and model release drama. I am spending more time using the tools deeply and watching where real leverage is emerging. The next wave of AI will not come from louder announcements. It will come from systems that actually do work over time.

Here is what I am most excited about.

Agentic workflows that actually replace work

Yes, agents still matter. But not the one turn wrappers or simple prompt chains we have all seen.

What I am watching closely is long running, multi step agentic workflows. Systems that can operate over hours or days, use browsers well, coordinate tasks, recover from failure, and complete objectives that either replace human activity or do things no human would ever do repeatedly at economic scale.

This is where implementation is finally catching up with ambition. We are moving out of the invention phase and into the phase where people are squeezing real productivity out of what already exists.

I love little tools like justcancel.io, but they are not changing our world.

Physical AI is arriving faster than most people expect

Physical AI is the other major breakout I am watching.

Self driving is real now. With Waymo expanding into major global cities, including London, we are going to see autonomous vehicles become normal far sooner than most expect. Within three years, this will feel transformational.

I desperately need this tech as I have 3 kids and loathe all the driving.

Robotics follows the same pattern. 2026 looks like another breakout year where AI finally leaves the screen and starts showing up in the physical world in meaningful ways. This is hard, capital intensive, and only a few teams can really push it forward. But when it moves, it moves fast.

Infrastructure keeps racing, even if consumers barely notice

In the background, the infrastructure race continues. Data centers, chips, power, and hardware are still fascinating and strategically critical.

From a consumer point of view, the impact is more subtle. Better products. Faster products. Cheaper products. That may not sound dramatic, but it is how every major technology wave compounds.

Most progress is implementation, not new algorithms

If you go deep into the weeds of AI right now, most teams are not inventing radically new models. They are trying to make what we already have actually work.

There are exceptions. Ilya Sutskever and his company Safe Superintelligence are clearly hunting for algorithmic breakthroughs beyond the transformer. You also hear about world models from researchers like Fei-Fei Li.

But for most of the ecosystem, the game is simpler and harder at the same time. Build agentic flows. Make them reliable. Integrate them into real workflows. Replace or augment real human activity.

That is where value is being created right now.

AI adoption is still a coordination problem

AI clearly threatens jobs. The evidence increasingly points that way. This affects new graduates, but it also affects senior roles.

At the same time, companies are hiring aggressively for AI product leaders. Accounting software firms, vertical SaaS companies, and all kinds of ISVs want leaders who can actually ship AI into products.

As Sarah Guo put it well, AI adoption has a coordination problem.

Using AI as an individual is obvious. Hundreds of millions of people use ChatGPT every week. But companies struggle. Roles change. Workflows change. Compliance matters. Compensation structures break. Org charts shift.

That is why replacing ten SDRs with AI is easy to describe and hard to execute at scale (100+ people). Small companies move first. Large companies move slower.

Creativity, video, and images keep improving

Video and image generation continue to get better at an impressive pace. The creative industries feel this first and most strongly. Other sectors will follow more slowly, but the tooling is clearly compounding.

Looking ahead

I am still loving this AI wave, but my attention is shifting from “efficiency” products to “abundance” products. Instead of just saving a few hours a week, how can I product improve the human condition?

In 2026, I am looking for deeper breakthroughs. Not just better text transformations, but systems that reason, act, and persist. Agentic systems that actually do work. Physical AI that shows up in the real world.

I am also watching what Elon Musk does next, especially with the massive compute cluster coming online in Memphis.

On this newsletter, I will keep writing about AI. Less news. More insight. More focus on the changes that actually shape where we are going.

If you have questions, hit reply. I read every message.

Let us see what 2026 brings.

Thanks for reading! 👋

If you liked this:

📨 Forward to a biz techie friend
💬 Hit reply and tell me what you want more (or less) of
📊Vote in the poll below 👇

— John
Prompt Punk