In 1979, Dan Bricklin and Bob Frankston launched VisiCalc, the first software designed for everyday users rather than engineers, hobbyists, or computer enthusiasts.

In 1982, another company inspired by VisiCalc released Multiplan, the predecessor of a product many of you are very used to and likely use daily. Arguably, one of the most successful software products ever. The company was called Microsoft, and Multiplan eventually became Excel.

Excel is just one example of what the software industry has delivered over the past 4 decades. When the personal computer appeared, most of the initial software was focused on productivity tools to make work easier and faster.

The distribution method has continually evolved since then, first with cassette tapes, then floppy disks, CDs, followed by internet downloads, and finally cloud-based software, which led to the so-called software-as-a-service (SaaS) business model, where software is charged on a subscription basis.

What hasn’t changed, especially in productivity and business software, is the need for user interaction. Modern tools, from HR and CRM systems to image editors and even code editors, are designed with the expectation that a human must operate them to get work done.

As AI and agents get more capable, many companies are transforming their products to go beyond just being a tool to do the actual work.

Everything is pointing towards a new world where products become hyper-personalized autonomous service providers, with a customization level even higher than current outsourcing agencies, just like if you had a dedicated team of tireless humans working 24/7, always available and with infinite memory, even surpassing the level of satisfaction you could have with a professional human-run agency today. All of that, while the new generation providers enjoy an infinite scaling opportunity.

Every company that grows eventually needs to manage people. Either outsourcing the process, losing some control, or building an in-house HR department, hiring some employees, and using SaaS tools to make their work easier.
In the coming years, a third option that merges the best of both will appear. You will subscribe to an HR tool that tailors itself to your company and handles the work autonomously.

Until now, designing creatives for an ad campaign required either buying and learning to use complex image editing tools or hiring a marketing agency or freelancer who is already an expert and provides faster and higher-quality work.
However, we already have tools where you upload a sketch, describe what you need, and for a few cents, you get the creative in different sizes and formats ready to be used. Eventually, as current models improve in reasoning capabilities, we will see marketing-focused platforms that serve as your marketing team, with significant cost reduction compared to today’s agencies and similar results.

Software development is one of the most affected industries. There are already hundreds, if not thousands, of coding agents in different formats. From IDE-integrated, like Cursor, Windsurf, Cline, Roo, … to platforms where you can describe your application and get it built and deployed, like Lovable, Bolt, Replit, … Businesses that need an application or website and have no technical team can now get it in minutes instead of weeks at one hundredth of the cost, with unlimited revisions and constant iteration and feature requests.
As development itself becomes trivial, most of the value is accruing at the infrastructure layer. Today’s state-of-the-art lacks capabilities to properly manage infrastructure autonomously. Still, the continued advancements suggest that autonomous infrastructure orchestration will also be solved sooner than later, closing the loop of application development and deployment.

In 2025, the global budget allocation for human amplification tooling was $295B. As the new generation software starts to take over both previous budgets and the headcount labor costs, there is an expected incremental opportunity in TAM for tools that actually work autonomously of $4.4T, as analyzed by McKinsey in Upgrading Software Business Models to Thrive in the AI Era. Reportedly, 46% of firms started to redirect headcount dollars into this new kind of intelligent software, looking for a 50-60% time savings without the proportional headcount costs.

Seat-based SaaS companies, with average margins of 62%, now need to figure out how to confront the agentic disruption from newcomers, which automate 70% of the tasks users need to perform manually. The new category reports even higher margins than traditional SaaS, expected to reach 80% as models optimize inference costs. The margin expansion comes from a shift in the pricing model, from seat-based to outcome-based. What previously could be an email sequencing tool for SDRs, where you pay per user, now takes care of everything from lead generation to closing the customer, and you pay per customer closed or a sales commission. This is what multiple startups refer to as an AI SDR.

Professional services and human-led agencies stand at the precipice. Just the subcategories for consulting, BPO, and outsourcing firms bill $1.5T annually on headcount arbitrage. They either adapt or die. The good part is that they have the opportunity to expand their margins to SaaS levels. They could already replace 66% of their current FTE (full-time equivalent) - the standardized total hours worked by all employees - a percentage that can only grow as AI continues improving.
Many of these service firms grow linearly, depending on their headcount. By integrating AI agents, it becomes non-linear. For a 500-person agency, this translates to replacing $10M in billable hours with $15M in agent-orchestrated deliverables.
Most of these service firms do not own a product or platform team capable of building the agents and infrastructure they need to survive, opening huge opportunities for startups. At the same time, they are the ones owning the customers, so if they adapt quickly, the opportunity is massive. That combination is leading to some startups being based on acquiring a firm, obtaining initial customers, and building the tech to scale the operations with AI agents.

A concern is how current workers will fare as AI takes over their workload. Some of them will switch roles, while others will evolve from manually serving a handful of customers to overseeing agents serving thousands.