The Coming Death of Closed SaaS – and What Comes After

01 /The Concept

Anyone who's sold a platform to large enterprises quickly discovers that every customer needs something slightly different – specific workflow support, specific integrations, specific behaviors. That creates a predictable set of problems:

- Some prospects walk away because what they need isn't there yet. - Some customers don't renew because the platform never fully matched their workflow. - Meanwhile, the engineering team is entirely consumed by one-off feature requests that may or may not actually close the deal.

Gigacatalyst is an AI tool designed to make platform customization faster and cheaper – per customer, per use case.

Setup involves connecting Gigacatalyst to your platform so it can learn the API and the design language of the UI. After that, product teams, salespeople, and even end users can add new features and automations by describing them in plain English. Gigacatalyst's AI programs them and integrates them directly into the platform interface as new commands and dashboards.

Custom features can be shared with other users – either across the entire platform or scoped to a specific company's team – through a built-in app-store-style marketplace.

The platform vendor pays for Gigacatalyst, with pricing based on AI usage volume. The startup claims it pays back within the first four weeks – through increased deal close rates and reduced churn.

Early results are encouraging: close rates up 31% on average, and the total contract value of renewals attributed to Gigacatalyst already exceeds $100,000.

The absolute numbers are still modest – Gigacatalyst is currently in Y Combinator, and the platform announcement went up on the YC site just a few days ago.

02 /Why It Matters

Gigacatalyst's founder described the underlying problem more sharply in a LinkedIn post.

It starts with a sales call where the prospect likes your product – but needs one specific feature before they'll sign. You check with engineering: two weeks to build and test. You sigh and tell the customer yes, because you need the revenue.

Then the same thing happens with the next prospect. And the next. Six months later, the engineering team isn't working on the product anymore. They've become an expensive custom development shop for your five biggest clients.

Five big clients is better than none. But it's not enough. And five clients means meaningful concentration risk – one departure is painful. Worse, these clients will eventually defect to a competitor anyway, because your product has stopped evolving.

A lot of people are currently saying "SaaS is dead." Gigacatalyst's founder has a more specific thesis: what's dead is the old conception of SaaS as a closed product. What's replacing it is SaaS as a *composable* platform – a base on which extensions, automations, and AI agents can be layered.

SaaS platforms will survive – but only the ones that make this layering easy, fast, cheap, and accessible to everyone: the vendor's own team, third-party developers, and the end users themselves.

In January this year, Legato raised $7 million in its first round ([related review](/review/prevrati-svoj-produkt-v-jekosistemu)) on a similar thesis, with specific emphasis on non-technical end users.

Legato's observation: "most business software isn't built for business users" Understandable, since most software is designed by engineers. Legato's goal is to let business users adapt and extend products around their own workflows – and share what they build through an app store of their own. That creates an ecosystem of business-specific applications around the core product, which drives broader distribution and higher revenue.

In more conceptual terms, Legato declared that the era of conventional partner ecosystems is giving way to "co-creation with users."

03 /Opportunities

Co-creation with users is a genuinely good idea. The problem is that most users are lazy. Getting people to build things is hard. Good idea, harsh reality.

Unless you account for something that's only becoming relevant now: soon, the primary users of business software won't be people – they'll be AI agents.

AI agents, unlike humans, aren't lazy. They're perfectly capable of building extensions, automations, and add-ons if given the right entry points, tools, and documentation. The co-creation idea might finally work – if you replace "users" with "AI agents"

That's the direction worth pursuing: SaaS built as composable platforms that AI agents can extend, adapt, and build on top of.

The architecture question almost answers itself: build the platform with an open extension model from day one, because the value of composability compounds with the number of AI agents that can build on top. The harder question is vertical selection – and the right answer is almost always the one you know best, not the one that looks biggest from the outside.