The Shrinking of Software: Why the Future of SaaS is Distributed

For twenty years, the story of SaaS has been one of expansion. Bigger platforms, more features, more integrations. Every quarter brought another product update, another tab, another reason to lock yourself into an ecosystem that promised to do everything.

AI has flipped that equation.

The cost of building software is decreasing. The speed to build has accelerated dramatically. What once required large teams and long development cycles can now be achieved faster, more intelligently, and at a fraction of the cost. The economics of creation have changed—and that changes everything about what software looks like and who builds it.

Growth Through Bloat

Historically, SaaS companies only had one direction to grow: outward.

To increase revenue, they expanded their feature sets. Innovation meant addition. "All in one" became the default business model.

That made sense when building software was slow and expensive. The result was a generation of platforms that could do almost anything—but that most people only ever used a small fraction of.

HubSpot and Salesforce are both incredible tools, but each carries the weight of success.

When building was hard, buying big made sense. When building becomes easy, it does not.

Distributed SaaS

AI is creating a new architecture for how software is built and consumed.

Instead of massive monoliths, we'll see ecosystems of small, intelligent microproducts—each doing one job brilliantly. Some will be built internally by companies to meet their exact needs. Others will be purchased off the shelf and connected through APIs or AI agents.

These microproducts will be threaded together dynamically, forming what I call Distributed SaaS—an ecosystem of modular tools that collectively deliver what used to come from a single platform.

Take Salesforce. Most companies use perhaps 30 or 40 percent of its capability: contact management, deal tracking, pipeline forecasting, and reporting. In a distributed world, those same companies could assemble their own stack with tools like Airtable or Folk for CRM structure, Clay for relationship enrichment, and an AI agent to handle lead routing and follow-ups. Add a lightweight dashboard for forecasting, and you have the 40 percent that actually matters—built faster, cheaper, and without the lock-in.

It's like replacing a Swiss Army knife with a collection of precision instruments—lighter, sharper, and infinitely easier to upgrade.

From Pyramids to Meshes

This shift changes how companies think about software assembly.

Tomorrow's tech stack won't be a pyramid with a few dominant tools at the base. It will be a mesh of interconnected products that can be swapped, combined, or extended without friction.

AI agents will increasingly manage this stitching—discovering, connecting, and maintaining the flow of data between microproducts. Internal teams will build only what's missing. Vendor relationships will move from multi-year lock-ins to short-term, high-value collaborations.

Software will become fluid.

What It Means for Builders and Buyers

For founders and product teams, this is both opportunity and risk.

Smaller products mean faster paths to market—but thinner moats. Defensibility will come not from the product itself but from the ecosystem it belongs to: the network effects, data flows, and AI orchestration layers that connect it.

For enterprises, flexibility will replace lock-in. Innovation cycles will shorten from years to weeks. Procurement will become experimentation. Instead of asking, "Which platform should we buy?", teams will ask, "Which combination of tools gets us there fastest?"

The power dynamic will shift—from providers to users.

The Unbundling of Software

We are moving from platforms to patterns, from static products to living systems.

The next decade won't be defined by who builds the biggest SaaS.It will be defined by who builds the smartest network of small, composable tools that act like one.

The future of software is distributed.Not because it's cheaper, but because intelligence moves to where it's needed most.

FAQ