One of the big upsides of AI Agents for knowledge work is the ROI changes dramatically on a number of things that you couldn’t have done before. There’s tons of work that we don’t do today because we can’t justify the “fixed cost” of getting it going. Almost every new idea becomes a meeting, with follow ups, and more coordination tax. So you, rightfully, prioritize only the highest impact work, and pray that you’ve made the right call on what that is. AI Agents changes the calculus here. The product team can afford to prototype more ideas to see which one is better. The business analyst can comb through more customer data to find a hidden insight. The engineer can build features faster. The legal team can better support smaller customers. The product marketer can run more campaigns or test more messages to reach more customers. Some of these things won’t matter a ton, of course. But many will. And by lowering the cost of trying a new idea, testing a marketing message, or researching a market, companies will start to do far more than before or at least get to their next destination faster.

The Figma case study is a great one. A team 100% focused on building great product, product led growth, fair pricing, building strong network effects, and expanding into logical adjacencies without getting too overextended. Many lessons for the next crop of startups.

Hard to overstate how slow the product design process was before Figma. You’d just brainstorm with low fi wireframes, feedback loops would take forever, and you’d wait weeks for mocks. Now you basically see a prototype of your product within hours of coming up with it.

Huge congrats to @zoink and the Figma team today. Figma revolutionized how we design and build software at Box - “low fidelity mock” is no longer a thing anymore. Watching Dylan and team for the last 12 years has been super fun, and can’t wait to see what they do next.

The TAM expansion that happens when software delivers the actual work output is bigger than we think. Software seats have been measured in the $100s per unit in a pre-agent world. But AI Agents’ work is getting measured in the $1,000s or $10,000s or more per unit.

AI Agents are going to blur the lines between the typical role definitions we’ve built up over the years. Any given function can now do more of its peer roles. The product manager can prototype a feature. The product marketer can create way more of their own marketing materials. The sales rep can generate their own product demos. And so on. At the same time, the higher output will causes new bottlenecks to emerge and new roles become needed to solve those. More product prototypes means more features will get the green light, and engineering tasks go up. More marketing materials getting created means more campaigns to run. Better sales demos means you can reach more customers and do more deals. And so on. This is what computers have always done to work. AI Agents just keeps this going. They make any role achieve more, which leads to more output. And output in one area creates new work in another area.

More has changed about how I think about work just in the last year than the past 20 years combined. As we’ve been going AI-first at Box, here are a few examples of what I run into. * Bias toward prototypes over a theory or pitch. My general expectation is that we should be seeing working examples of what’s possible or what we’re building far more than just talking about it or just jumping on a white board. * Most projects are “going bigger”. In a world where you can design more, build more, and produce more content, there’s almost no scenario where you end up thinking small in a project. This is happening in marketing, product, sales, and more. Counterintuitively (to some), by AI allowing us to go bigger, we have actually hired in new roles we wouldn’t have justified before. * The lines between classically defined roles are likely going to blur more and more. A designer can now be thinking more like a PM. A product marketer can do much more of the surrounding marketing functions and more. It’s sort of a race to see which roles bleed into others, and then what new types of roles emerge out of that in the process. * I expect way more research to be done for almost any topic. I find myself asking “have we checked how the top 20 software companies are doing X?” in a ton of meetings. The expectation is that we should know vastly more about *everything*. This doesn’t change judgment and intuition, just adds more insight in the process. * I will send off tasks to AI before I bug someone else on something relatively straightforward. If I need a piece of information in the business I can ask Box AI. If I need to research a topic, I can send off a deep research task to an agent. If I want to see a mock myself I can have AI design a quick example. The aggregate result of all of this is just the expectation that everything should move faster and output should go up. We’ve effectively expanded the capabilities of the company overnight. It will be wild to see what this looks like in 5 years from now.

Work is normally rate limited by how fast we can individually execute. AI flips this and turns us all into managers of agents. In this world, a lot of work comes down to coming up with the best prompts for agents, reviewing their work, and integrating their output.

The trillion dollar opportunity in enterprise software is AI Agents. The reason for this is AI Agents expand many software categories because most tools have been constrained by the number of users on the other end of the tool. Enterprise software traditionally enabled people to do their work. But now the software also comes with actual productive output as well. This then breaks the traditional limits that many software markets have had, because smaller customers can use these tools more, new departments and industries open up, and previously scarce areas of work can be scaled more. For instance, most estimates put the code IDE category at a few billion dollars just a few years ago. This has - amazingly - always been a very small category of software. Well, now with AI Agents, the IDE market and coding in general is one of the fastest categories of spend in tech. This is because it’s bringing automation to a high value activity and supplying the world with a traditionally very scarce and high cost resource, engineers. The same is going to be true for many other categories of work across legal, financial services, healthcare, and more. Amazingly, this now opens up many niche categories that just wouldn’t have been economically viable for a software company before. For instance, making a business out of selling software to life sciences regulatory compliance managers would have been sub-scale before, but in a world of AI Agents this all of a sudden becomes possible. We’re already seeing this in our initial Box AI Agent use cases. Many of the early Agents being created by customers are for automating or augmenting work in previously underserved areas. This is letting customers automate a process they never would have gotten around to before, or dramatically expanding the output of work they were already doing. In all cases it’s more TAM that software wasn’t touching before. In all, it’s clear that AI Agents are going to grow many software categories. It’s an incredible time to be going after these spaces because even the small ones will now be massive, and the big ones will just get bigger. Tons of opportunity.

AI Agents expand many software categories because most tools have been constrained by the number of potential users on the other end. These systems now deliver work, which means you have uncapped usage potential. Plus, any size company and new departments become buyers.