Is AI the end of SaaS as we know it?

Shares in several prominent public SaaS companies fell approximately 10% heading into February, wiping out $300 billion in value. One of the instigators was Anthropic’s release of Claude Cowork with plugins in legal, marketing, and other business functions. Investors also read how AI writes 90% of Claude’s code, while technologists noticed that Claude Opus 4.5 and GPT-5.2 Codex hit 80%+ on the SWE-bench Verified benchmark.

Watching the news on those days, I heard financial analysts suggest that enterprises might walk away from their expensive SaaS contracts and replace them with AI capabilities, including in-house developed AI agents.

While a collapse in the SaaS business model seems improbable, recent advances in AI are a paradigm shift in how organizations develop and procure software capabilities. Last year, I asked whether AI is the end of IT as we know it and provided suggestions for what world-class IT looks like in the AI era. While markets have stabilized somewhat, CIOs must consider whether AI is the end of SaaS as we know it.  

“I don’t see AI agents taking over the functions of traditional SaaS apps, but they are currently redefining the role of software in the enterprise,” says Udo Sglavo, VP of applied AI and modeling at SAS. “Agents will increasingly act as autonomous operators across systems, abstracting away traditional graphical user interfaces and exposing capabilities through natural language and intent. That doesn’t mean SaaS will disappear; it will just become more composable, embedded, and invisible.”

AI’s software development capabilities and its ability to gain business function expertise raise several questions. Should CIO’s build more and buy fewer new SaaS capabilities? When should CIOs renew SaaS contracts rather than consolidate their existing SaaS investments? Below are five questions worth reviewing.

When should vibe coding replace SaaS?

Top CIOs will embrace AI’s democratization as the evolution of no-code and low-code development capabilities. Many will experiment with vibe coding to learn where they are effective in prototyping and whether they can deliver capabilities that meet their release criteria for AI agents.

A simpler, lower-cost build option may be palatable to some CIOs facing SaaS sprawl and seeking a more integrated user experience. For example, there are more than 15,000 marketing solutions, where integrating customer data and multiple SaaS solutions can be challenging.

There may also be cost considerations. Across all departments, large enterprises average more than 600 SaaS applications and spend $280 million annually on SaaS.

AI isn’t just coding; it can also help write requirements, maintain documentation, and simplify some multicloud complexities. The 2026 Agentic Coding Trends Report by Anthropic states, “Engineering teams discovered that AI can now handle entire implementation workflows: writing tests, debugging failures, generating documentation, and navigating increasingly complex codebases.”

So, should CIOs build more AI agents and applications because of AI’s coding capabilities, rather than buy or retain SaaS solutions?

Some CIOs may dismiss AI’s development capabilities as overly hyped or point to the added work (and cost) for evaluating whether an AI’s code meets coding requirements. More optimistic CIOs will seek ways to leverage AI code-generating tools to simplify the tech stack, reduce costs, or accelerate innovation. 

“Building makes sense when you have proprietary data or workflows no vendor can replicate, but only when your delivery pipelines, automated testing, and security controls can match the pace AI generates code,” says Ashwin Mithra, global head of information security at CloudBees. “Without that foundation, you’re not innovating faster; you’re compounding risks you can’t yet see.”

Before shifting to a build mentality, CIOs should evaluate whether their teams have the practices to support the full software development lifecycle (SDLC) and whether there’s sufficient funding for it. CIOs thinking about building more than they’ve done in the past should ask:

“The latest models make it easier for teams to prototype and build targeted tools quickly,” says Priya Vijayarajendran, CEO at ASAPP. “But most enterprise value comes from systems that can execute reliably across complex environments, with governance, integrations, and auditability built into the platform.”

What AI capabilities should I add to my SaaS evaluation?

I suspect far fewer CIOs will shift to building AI agents just because coding is easier. That said, CIOs should also raise the bar for the AI capabilities they expect from existing SaaS solutions and those under evaluation.

Experts suggest extending the enterprise evaluation around SaaS applications to include AI criteria, providing several examples:

  • Srikrishnan Ganesan, co-founder and CEO at Rocketlane, suggests evaluating whether the product offers the most ambitious AI capabilities in its category and whether you can extend them with vibe coding to ease the building and maintenance of custom applications. 
  • Titus Capilnean, VP at Civic Technologies, says employees are already using agents to access SaaS tools, so CIOs need to consider agent auditability, access governance, revocation, and how it fits into their infrastructure.
  • Ronak Sheth, CEO at Pricefx, says to look for providers that embed governed, explainable AI directly into core business workflows and that orchestrate decisions across systems of authority, not those that layer chat on top of legacy screens.

When a SaaS platform is strategic and deeply embedded in existing operations, CIOs should review its product roadmaps, send staff members to its conferences, invest in training, and become active in customer advisory boards.

What should I expect around SaaS pricing?

CIOs are unlikely to abandon SaaS providers as fast as some investors sell their stocks. It’s also unlikely that SaaS companies will pass AI’s productivity gains on to customers in the form of lower prices.

“AI is materially changing development economics; the productivity gains are real, but the assumption that efficiency gains flow automatically to customers underestimates how SaaS pricing actually works,” says Bryan Gibson, VP of operational excellence at Teramind. “What CIOs should push for is more value at the same price point, not necessarily lower prices. The right pressure to apply at renewal is: Show me what AI has changed about your platform’s capability and my team’s productivity, and if the answer is thin, that’s negotiating leverage.”

Others suggest reviewing how SaaS companies are adjusting their pricing structures as work shifts from people to AI agents.

“CIOs should evaluate whether their SaaS providers are evolving beyond seat-based pricing toward outcome-based value,” says Adi Kuruganti, chief AI and development officer at Automation Anywhere. “Traditional SaaS models assume humans execute workflows, but AI agents increasingly perform work autonomously. If pricing is still tied to user access instead of measurable results, that’s a warning sign.”

If businesses continue to lay off people driven by AI productivity gains, it will affect SaaS companies with per-user licensing and subscription models. CIOs should expand their FinOps responsibilities to track shifts from per-user to other usage-based pricing models and their associated cost implications.

How should I prepare my organization for workflow changes?

One certainty is that CIOs must prep their organization for workflow changes. There are three scenarios to consider that all require change management efforts:

  • Building internal capabilities to replace SaaS or consolidating SaaS platforms will need a change management plan to support the transition.
  • Potential SaaS market impacts, such as provider acquisitions, reduced support offerings, or vendor closings, will require CIOs to develop risk mitigation and change plans.
  • As SaaS solutions offer more AI agents with increased sophistication, ensuring adoption of those capabilities will also require change management efforts.

“The future of some SaaS platforms is not safe, just as the future of hardware, foundational models, and the next great innovation we haven’t yet conceived of will not be safe,” says Piyush Patel, chief ecosystem officer at Algolia. “CIOs must prepare their teams for massive SaaS changes by focusing on only one objective: delivering value to customers.”

Where should I seek differentiating capabilities?

One final question is how SaaS platforms market their differentiating capabilities. CIOs should look for these value adds that are not easy to quantify in build-versus-buy decisions or when comparing alternative solutions.

“There’s an uncomfortable question confronting those of us who sell software for a living: If everyone can build, why would anyone buy?” says Chuck Ganapathi, CEO at Gainsight. “The moat isn’t code anymore — it’s context, domain expertise, and the trust you gain from serving thousands of customers for decades.”

One important area SaaS platforms need to improve is their data strategy. Providing forms, reports, and workflows against siloed data sources is no longer viable. SaaS must demonstrate that it delivers value using the enterprise’s proprietary data, wherever it resides, and in accordance with the company’s data governance and compliance requirements.

“The winners in an AI-driven world apply AI to proprietary data sets that continually update, and embed AI into existing workflows so adoption doesn’t require changing everything,” says Jaime Meritt, chief product officer at Verint. “They will provide deep domain knowledge baked into every model, bot, and automation.”

CIOs should also review their SaaS platforms’ AI capabilities and ensure they enable multi-agent workflows with support for MCP server integrations.

“The warning sign for any SaaS company is whether it exposes an MCP server,”  says Andy Berman, CEO and co-founder at Runlayer. “The ones that become agent-accessible infrastructure survive, while the others clinging to human-only UIs are already being routed around.”

The challenging part for CIOs is that AI models and agentic AI capabilities continue to evolve. Today’s build-versus-buy decisions need to be future-proofed, but it’s highly unlikely we’ll see the end of SaaS because of AI.