AI agents are learning to reverse-engineer existing software and recreate its features without access to the original source code. If that becomes cheap and reliable enough, software companies could lose the advantage that lets them charge customers recurring fees for products that took years to develop.
REA (Reverse Engineer Anything) gives AI agents tools to inspect applications, binaries and other software. It does not automatically reproduce an entire application, but it lowers the barrier to understanding how existing products work.
https://github.com/morluto/rea
The threat to Adobe, Microsoft and smaller software vendors is straightforward: customers may no longer need an entire subscription to get the features they actually use. An AI-built alternative that performs a few essential tasks could be enough to cancel a costly plan or negotiate a lower price.
A clone does not have to match the original product feature for feature. It only has to be good enough for a customer to stop paying.
That creates a new problem for software companies. Their development costs may remain high even as competitors find it cheaper to reproduce their functionality.
Google could benefit from that shift. It owns Android, Chrome, Search, Gemini and Google Cloud. If developers create more software with AI, they still need models, computing infrastructure and ways to distribute their products.
But Google faces a contradiction: AI agents could also reduce dependence on traditional search. If users ask an agent for answers and completed tasks instead of clicking through search results, Google’s advertising business could come under pressure.
Google is trying to capture the new interface. On October 8, 2026, it introduced a Gemini agent designed to carry out workplace tasks across applications. The strategy is to move beyond answering questions toward coordinating the tools people use to get work done.
The investment question is whether Google can earn more from AI agents and cloud services than it loses if AI changes how people use Search and other existing products.
The reported 43% odds of Google leading Anthropic in the AI race by year-end do not answer that question. Model rankings are not profit forecasts. AI usage can grow while prices fall and computing costs remain high.
The deeper risk for software companies is that customers gain bargaining power when reproducing features becomes cheaper. The opportunity for Google is to supply the infrastructure and agent layer that makes those alternatives possible.
The danger is that Google supplies the infrastructure while a competitor owns the customer.
That is the real contest: not who can clone the most software, but who collects the money after software becomes easier to reproduce.
Not financial advice.
Got a news tip or correction? Let us know
If you got something out of this, please chip in to keep this site running, or subscribe to go ad-free.
43 views