Frontier Model Developers Ride High, But SaaS Vendors Remain Critical To Enterprise AI Value Creation

Frontier AI developers are riding a wave of momentum, with rapid innovation and IPO expectations continuing to capture buyer attention. Increasingly, enterprises are incorporating Anthropic, OpenAI and similar providers into procurement exercises, using them as a credible alternative pathway to traditional software by:

1. Building AI applications through natural language with coding agent platforms (such as Claude Code, Codex and Cursor) and enterprise AI development platforms (such as OpenAI Frontier). Business users face much lower barriers to entry to develop applications and connect them directly to enterprise systems via standards such as model context protocol (MCP). This accelerates development cycles significantly, while enabling organizations to own and control their applications.
2. Buying verticalized AI agents off-the-shelf from frontier model labs that have built in-house products (for example, legal or financial services workflows) and co-developed solutions alongside systems integrators and consultants.

At the same time, frontier AI developers are moving to the application layer. Anthropic and OpenAI are building partnerships with systems integrators and consultants, while also developing verticalized, ready-to-deploy agents.

Given these options, enterprises are increasingly questioning whether it remains sensible to pay a premium for native AI capabilities embedded within SaaS platforms.

Nevertheless, SaaS vendors retain several structural advantages that enterprise buyers should carefully weigh during procurement decisions, namely:

• Deep domain expertise: While frontier models are increasingly capable within specific industries, building enterprise-grade applications requires more than model capability. The Verdantix 2025 Global Corporate AI Survey shows that ‘industry expertise and existing customer base’ is the second-highest priority criteria for enterprise when selecting AI-enabled software. SMEs have a deep understanding of industry-specific processes, integration complexity and how AI can deliver measurable outcomes for end-users in different roles. This is especially important in highly regulated industries with strong governance requirements.
• Embedded workflows: SaaS platforms sit at the centre of enterprise operations, with workflows deeply integrated into daily processes. AI functionality built from the ground up by SaaS vendors can be more effectively integrated into processes, reducing friction and accelerating adoption – especially for front-line workers. Relegating critical software applications to systems of record risks disrupting processes that enterprises rely on, and disenfranchising end-users.
• Proprietary data advantage: High-quality data are essential for effective enterprise AI agent implementation – whether for post-training, fine-tuning, contextual reasoning or building deterministic workflows using knowledge graphs. SaaS vendors possess years of structured and operational data, alongside access to domain experts. This combination is difficult for frontier model developers to replicate at scale.
• Model flexibility: Frontier model developers are inherently tied to their own models. By contrast, SaaS vendors can abstract the model layer – selecting or recommending the best model for each use case based on cost, performance and risk. This flexibility mitigates vendor lock-in and protects against issues such as token inflation or platform-specific disruptions (witness the US government effectively banning Anthropic’s Mythos and Fable models).

Ultimately, the real differentiator for SaaS comes down to execution. In enterprise environments, the quality of the underlying model is not the be-all and end-all: the focus instead is on reliable, auditable and value-driven AI workflows with deep integration.

SaaS vendors often excel in this ‘engineering discipline’, building robust systems that combine domain knowledge, governance and workflow integration to deliver consistent outcomes. That said, they cannot afford complacency. To remain competitive, they must:

1) Invest aggressively in differentiated, customer-aligned AI capabilities.

2) Embed AI deeply into existing workflows, creating frictionless experiences and driving customer stickiness.

3) Focus on delivering immediate, measurable value to customers.

For more insights into the impact of AI agents on enterprise software, please read Verdantix Strategic Focus: How AI Changes Enterprise Software Business Cases.