Editor's Note: In the past, there was a clear path for enterprise software startups: first, find a niche entry point small enough but with growth potential, and achieve tens of millions of dollars in ARR with a single-point product (niche feature entry); then, expand the product suite around the same buyer to drive revenue towards hundreds of millions of dollars (expand into a product suite); finally, with sufficient user and data accumulation, become a new platform (reshape the underlying platform).

But in the AI era, this "Three-Act Play" is becoming obsolete. As software development costs plummet and the cycle from conception to launch is drastically compressed, startups no longer need to spend three to five years validating a niche market before slowly expanding their boundaries. Companies like Cursor, Clay, and Harvey have gone from zero to approaching or even surpassing $100 million ARR in a short time, indicating that the competitive pace of enterprise software has been rewritten.

The core thesis of this article is: In a rapidly changing market, relying on a "safe wedge" might actually become conservative. The new generation of software companies needs not just to find a wedge, but to possess, from the outset, the ambition to reconstruct entire workflows or even replace existing platforms. The so-called death of the "Three-Act Play" is, in essence, the beginning of a shift from incremental expansion to going all-in from the start.

The following is the original text:

In the past, building an enterprise software company had a fairly clear playbook.

Act I: The Wedge, or Unbundling

Start by picking off a function or market segment underserved by existing solutions. During a platform shift, you'd take a function from the incumbent platform and make it 10x better under the new paradigm, using that as your entry wedge.

This segment had to be big enough to get a company to tens of millions of dollars in ARR quickly, but not so big that it invited crushing competition immediately. Statsig started with product experimentation; Rippling started with employee onboarding/offboarding orchestration, and so on.

Most startups would spend 3-5 years iterating on the initial product, building out an early GTM motion, and scaling to $10-$50M ARR before moving to Act II.

Act II: The Suite

The core of Act II was to launch adjacent products to get the company through the $100M ARR mark. Here, you were no longer just a single-point product, but started to build out a portfolio.

Statsig started with product experiments, then added feature flags, session replay, product analytics, etc. Rippling started with payroll/HR workflows, i.e., onboarding/offboarding, and then filled out a suite of HR, benefits, recruiting products, etc., all sold to the same buyer.

For companies that made it this far, this usually took another 3-5 years in calendar time. As the first product scaled to ~$50M ARR, the company began cross-selling the 2nd and 3rd products. By $100M ARR, maybe the next two products were at $10M and $1M ARR respectively. It was this suite play that opened the path to $200M, $500M ARR and beyond.

Act III: The Platform

The final stage was rebundling. As the company amassed enough scale and user engagement, you'd eventually earn the right to replace the underlying platform you were built on. This was the basic logic of all Systems of Engagement attempting to commoditize their underlying Systems of Record. In theory, this was the path to $5B+ in durable, sticky revenue.

Speedrunning the Playbook

I'm worried this three-act play is dead. I think the world is moving too fast now.

The three-act path implicitly relied on a certain amount of calendar time, especially in the early days. Founders could only do so much: first, focus on finding product-market fit, then build an early GTM motion, then scale GTM. The reason you wouldn't start Act II before getting to $10-$50M ARR was that you were still putting all your energy into Act I.

In the last couple of years, we've seen a cohort of companies go from near 0 to $100M ARR, e.g., Cursor, Cognition, Clay, Harvey, Sierra, Baseten, Fireworks, Lovable. This, in itself, is evidence that the world has already changed.

There's no time left to be too precious about the step-by-step strategy anymore. As the cost of software engineering plummets, the time required to complete Act I and Act II converges to zero. I think the rational thing to do now is to plan on building everything fast, from the start.

Ambition

This also deeply changes how I think about early-stage investing. In the past, I looked for a protective wedge—a harbor where a company could safely get to $10-$50M ARR. Now, the wedge almost feels too small. I find myself wanting founders to jump into the deep end.

For example, I remember meeting Anysphere, aka Cursor, at the seed stage. Their plan, it seemed, was to just replace VS Code because they thought VS Code was too limiting for AI programming. I thought that was crazy at the time—VS Code was very loved. After years of IDE fragmentation, VS Code had finally won. Why would a seed-stage company try to replace VS Code from the start? The more sensible path seemed to be to build a plugin first, and then earn the right to replace it.

I was wrong, by the way. In hindsight, replacing VS Code doesn't even seem ambitious enough. Why stop there?

As the cost of writing software converges to zero, I find myself caring more about ambition than anything else. Not normal ambition, but unreasonable, relentless ambition.

I think the three-act play is over. In a period of rapid change, depending on a wedge is too timid. If you're going to do it, you might as well go for the whole thing from the start.