Redundancies. Strategy shifts. AI-driven change. In this episode of the Dootrix podcast, Rob and Kevin take a hard look at what Microsoft’s latest job cuts signal. This is not just about big tech. It is about the wider world of work.
Since 2023, Microsoft has cut over 24,000 roles, including 9,000 in their most recent round. On the surface, these cuts appear strategic rather than financial. Microsoft’s revenues are going up, not down. The layoffs coincided with an internal memo leaked by the president of their developer division. It confirmed that AI usage would become a tracked metric in employee reviews. AI is no longer optional. It is now central to how work gets done, and increasingly, who gets to do the work at all.
This is not just about making people more productive. It is about removing the work altogether.
Rob points to Microsoft’s phrasing. The company said it was focused on “ending or decreasing work.” That signals a shift from enabling people to do more, to replacing people with tools that can do the job themselves. The rise of agentic AI is making this possible. These systems do not just assist with tasks. They achieve outcomes on your behalf.
Kevin suggests this shift may be felt first in large organisations. Microsoft’s sales and account teams used to operate in packs, with multiple specialists assigned to each account. But if AI can be part of that team, some of those roles disappear. In development, senior engineers will spend less time delegating to junior staff and more time orchestrating agents.
Teams are being hollowed out. Consultancy firms are already producing more work with fewer people. Clients still pay the same rates, but much of the work is now delivered by AI.
Rob explains how roles are being unbundled. The value lies in deciding what needs to be done. That part still needs human judgment. But everything that comes after, the work of turning intent into execution, is now in reach of AI. This is where the tooling excels.
He shares an example from a Dootrix architect. Faced with a legacy codebase, the architect needed to build a test harness to explore a theory. In the past, this would have taken time or required support from junior engineers. Instead, he wrote a prompt. The test harness and data were generated automatically, proving or disproving his theory in minutes. That freed him up to focus elsewhere. The result was the same, but with less friction and no extra hands.
Kevin raises a problem. Despite all the tools, he feels busier than ever. AI was supposed to reduce the amount of work. In practice, it has increased the volume and velocity of tasks. Expectations are rising. The pace is accelerating. And there is a deeper risk beneath it.
If AI eliminates junior roles, where will the future seniors come from? Today’s efficiency could create tomorrow’s skills shortage. The learning curve might collapse before new talent has a chance to climb it.
We have seen this before. Spreadsheets and productivity apps promised to save time. Instead, they allowed us to cram more into the same day. Allegedly, this is good for the economy. But inflation is rising and most people feel worse off.
This shift will not just change how work is done. It will reshape what people expect from the products they use. If AI is delivering results at work, consumers will demand the same from the tools in their lives.
Kevin shares his recent experience with an appraisal tool. He used AI to turn his raw thoughts into a structured review. But he then had to paste that output into a rigid SaaS form. The process felt backward. The work had already been done. The software was now just friction.
Rob highlights a wider issue. Many SaaS tools are just form wrappers. That model no longer works. Products need to evolve. The interface is becoming an obstacle. The value lies in what the tool enables, not how many fields it captures.
This leads to a clear warning for product teams. If your product is essentially a form, it is time to rethink. Rob believes that 99 percent of SaaS tools will need to be redesigned. The world is changing around them. The workplace is evolving, and customers will not tolerate unnecessary effort for long.
Kevin imagines a world where interfaces are created on demand. LLMs could generate full applications as needed. You would not browse apps. You would describe what you want, and it would appear. This is not a distant dream. It is the beginning of a shift toward invisible products and ambient computing.
The episode then explores where AI is heading next. Kevin introduces JEPA, a new model architecture developed by Meta. JEPA is trained on images and video, rather than just text. It learns not by predicting pixels, but by understanding the relationships between objects in space. This gives it a way of grasping physical reality.
Where LLMs understand language, JEPA begins to understand the physical world. This could be a step toward superintelligence. AI that does not just repeat knowledge, but begins to create new science.
Kevin also discusses Centure, a model trained to predict human decisions based on psychological experiments. By fine-tuning an LLM on over 60,000 responses, researchers achieved a high level of predictive accuracy. The fact that language models can be shaped to understand human behaviour opens up new territory. It also blurs the boundary between understanding and influence.
Finally, Rob highlights an ongoing legal battle in the United States. Anthropic was found to have trained its AI on thousands of copyrighted books. A judge ruled that this was not a breach of copyright. The reason? The use was “exceedingly transformative.” The books were not copied. They were used to create something new.
This opens the door for AI companies to continue training on copyrighted material without payment. Meanwhile, Cloudflare has launched a tool to block AI crawlers. Their argument is not about legality. It is about fairness. If content is used to train a model, the creators should benefit from the reach it enables.
As Rob points out, this is no longer a copyright debate. It is a question of control. The AI companies want reach. The creators want recognition. And the technology is moving faster than the rules that govern it.
The future of work is not simply a productivity upgrade. It is a complete rewrite. Leaders, developers, and product teams all need to pay attention. The shift is underway. The question is whether we are ready for it.