It’s crazy how much AI Agents are going to change the core premise of software.
Since the beginning of the industry, software has been something that enables people to do their work. We’ve spent decades building features that make people incrementally more productive when using the technology. Every feature has been in service of extracting a few percentage points of productivity improvement.
But no matter how many features get built, software before AI was ultimately still bound by how fast the human using the tool was able to do their work. Word processing or coding was always constrained by the upper limit of how fast you could type what you wanted. Contract software was constrained by how quickly you could review a document. Sales tools were limited by your ability to add, review, or interact with clients.
AI Agents flips this on its head. The new paradigm of software will soon be about AI Agents doing work for you, with humans used to plan, review and orchestration step in the process.
Think about the impact of this. Now, all of software is effectively unbounded by the speed at which humans work. Or, better yet, even their availability in the first place.
You can already see how this workflow is shaping up, mostly notably starting with AI coding. The new AI coding workflow often is to have an AI Agent generate some small or large chunk of code based on your prompt, and then you go through and review its work, often making a few changes and fixes in the process. In most cases this accelerates what you’re building meaningfully. Of course there still are plenty of bottlenecks elsewhere in the process, so you don’t capture all the productivity improvement, but you get a major jump. And these are just the early days.
We’re also starting to see similar output improvements in other areas of work. Deep research will take work that used to take a few hours or days for an analyst and compress it into minutes. AI Agents that review contracts or generate proposals equally compress this much time as well. The workflows are all the same: send a task to an AI Agent, review the output and incorporate it into your work.
The biggest impact is how we will need to think about and design software in the future. The tools that start to emerge (or that are reinvented) are no longer focused on how humans can become incrementally more productive in our work; instead, tools will be optimized for how we interact with AI Agents to make *them* orders of magnitude more productive and effective. This is about software the supports the planning, review, orchestration, and consumption of AI Agents’ work.
This is truly a new paradigm of software that’s emerging.
Ovo što se dešava je čist primer zakona privlačnosti. Onaj čije se ime ne pominje je toliko govorio o Narodnom Pokretu i Svemir mu je dao Narodni Pokret.
Rebecca Lovell and Megan Lovell of #LarkinPoe, winners of the "Best Contemporary Blues Album" award for "Blood Harmony"
Photo by Emma McIntyre/Getty Images