CIO.com has pushed back on one of the loudest takes in enterprise software: the claim, popularized over the past two years by venture voices arguing that AI agents will bypass application layers entirely, that SaaS as a category is finished. The publication's counterargument is narrower and more useful for product teams: SaaS isn't dead, but the market is restructuring into something more hybrid.

That distinction matters for roadmaps. "SaaS is dead" is a categorical prediction; "the market is becoming hybrid" is a structural description of what buyers are already doing. If the second framing is right, product managers don't face a cliff where subscription software disappears. They face a gradual unbundling of how value gets delivered — some of it through full applications, some through APIs, some through AI-mediated access to the same underlying capability.

What does "hybrid" mean concretely here? The CIO.com piece points to a market where traditional multi-tenant SaaS subscriptions coexist with other consumption patterns rather than being replaced by them. For enterprise buyers, that means portfolios increasingly mix conventional SaaS platforms, platform-as-a-service components, and AI-driven interfaces that sit on top of — or alongside — established applications. For product managers, that means the monolithic SaaS bundle is no longer the only assumed unit of value.

The implications for pricing and packaging are immediate. If customers can reach your product's capability through an API call orchestrated by an agent, or through a traditional seat-based UI, you need packaging that works across both. Usage-based pricing, priced against metered consumption, becomes less a growth-stage fashion and more a defensive necessity — otherwise an intermediary captures the transaction and your seat count quietly erodes. Pricing teams should be stress-testing what happens to revenue when a customer's "seats" are effectively AI agents rather than people logging in.

There are failure modes to keep honest about. Hybrid markets reward incumbents with strong data positions and punish vendors whose differentiation lives entirely in workflow UI, since UI is precisely what AI interfaces can replicate or bypass. Teams with thin proprietary data and a front-end-heavy value proposition face genuine displacement risk — hybrid doesn't rescue everyone. And the transition period is messy: dual delivery models raise support costs, complicate compliance boundaries, and can confuse sales compensation if packaging splits between seats and consumption.

The conditions under which the hybrid thesis holds also deserve scrutiny. It assumes buyers still want governed, vendor-maintained software for core systems of record while experimenting with AI-mediated access at the edges. If that assumption breaks — if agents prove reliable enough to run core workflows against raw data stores without application logic in between — hybrid becomes a waypoint, not a destination. The CIO.com argument is credible for the near term precisely because enterprise procurement, security review, and integration debt all change slowly, not because the technology can't move faster.

For product managers planning next quarter, the practical read is this: don't rip up your SaaS roadmap, but stop assuming the application interface is the only front door. Instrument API usage separately from UI usage, map which of your workflows an agent could plausibly execute end-to-end today, and identify where your proprietary data — not your screens — is the actual moat.

The debate over SaaS's demise will keep generating headlines, but the operational reality for most product orgs in 2025 and beyond is blending delivery models, not wholesale replacement — and the teams that instrument and price for both paths early will set the terms for the next procurement cycle.