Reclaiming SaaS Value in the Agentic AI Era

By Jacques Bughin 

As agentic AI takes over task execution, the foundations of traditional SaaS begin to crumble. Interfaces lose value, orchestration gains power, and control over workflows becomes the new battleground. Dr. Jacques Bughin reveals how companies must pivot fast or risk fading into irrelevance in this rapidly shifting landscape.

1. The End of SaaS as We Know It

The traditional Software as a Service (SaaS) model, characterized by user interfaces, per-seat pricing, and feature sets, is undergoing a significant transformation. The catalyst for this change is the emergence of agentic AI—autonomous digital workers capable of executing tasks on behalf of users. By 2030, it’s projected that 30% of current B2B SaaS revenue is at risk due to orchestration-driven compression. Users will delegate tasks to agents, potentially bypassing multiple software interfaces. This shift changes the pricing paradigm from “software access” to “outcome fulfillment.. SaaS companies, especially those offering horizontal, mid-layer, or UI-heavy solutions, face significant disruption. Products like dashboards, CRMs, schedulers, or project tools without vertical integration risk obsolescence within 2–4 years

Who Owns the Stack Now and how it will change

In the agentic era, value accrues to those controlling the stack: infrastructure providers like Azure, AWS, and GCP; model developers such as OpenAI, Claude, Gemini, and Mistral; orchestrators including Copilot, Gemini 1.5, Dust, and LangGraph; and finally, the SaaS layer, which is increasingly reduced to an API endpoint. Control over the user interface no longer ensures monetization. Agents are UX-agnostic; what matters is control over intent, memory, context, and orchestration flow. The entity that owns the agent effectively owns the user.

Traditional SaaS captures value through accounts, permissions, UIs, and reports. Agentic AI, however, derives value from workflow dominance, automation logic, and autonomous action. This represents a shift from user-driven interfaces to autonomous decision-making architectures. Instead of users navigating multiple tools, agents pull data, query APIs, make decisions, and inform users, rendering entire layers of the traditional stack obsolete.

By the way, the transition is accelerating:

• Adobe has integrated agentic AI into its suite, enhancing user experiences across its platforms. Salesforce has expanded its family of large action models, designed to predict and perform next actions, powering AI agents across its ecosystem. ServiceNow has introduced autonomous AI agents capable of executing complex tasks, differentiating from traditional generative AI copilots. Startups like Sana (SE), Otherside AI, and Deepop are building orchestrators as core products. Additionally, tools like AutoGen, CrewAI, and LangGraph are rapidly maturing, facilitating agent deployment
• Major tech companies are aligning around task flow control. Microsoft is transforming M365 into an orchestration hub via Copilot. Google utilizes Gemini to defend search and expand SaaS offerings. Amazon connects agents to transactions through Bedrock and Alexa. Meta develops social/consumer agents to protect advertising revenue. NVIDIA and AMD drive demand for compute through orchestration. Each stands to gain from cloud revenue, LLM licensing, chip sales, or agent UX control.

In parallel, the cost per 1,000 tokens for LLMs has plummeted from ~$10 (GPT-4, 2023) to less than $0.01. Task orchestration costs are decreasing by 80–90% annually, while the number of addressable agentic workflows grows exponentially. By 2026, many SaaS workflows will be more cost-effective and efficient when executed by agentic systems rather than traditional applications.

2. SAAS future is different

Cutting Costs Isn’t a strategy, you need to move

Some SaaS firms may respond by reducing R&D, narrowing product scope, or halting innovation. While this might preserve short-term EBITDA, it jeopardizes long-term viability. Agents will outperform pared-down tools, leading to user attrition and increased churn. Eventually, these firms risk becoming mere wrappers around others’ orchestration logic.

Reality bites: Klarna, a fintech leader, has undertaken a significant transformation by reducing its reliance on over 1,200 SaaS tools, opting for internally developed AI-powered solutions. This strategic shift led to annual savings exceeding $10 million and streamlined operations. Notably, Klarna severed ties with major SaaS providers like Salesforce and Workday, replacing them with internal systems built on AI infrastructure, including OpenAI’s technologies. The company’s AI assistant, powered by OpenAI, managed two-thirds of customer service chats in its first month, effectively performing the work of 700 full-time agents. This move not only improved efficiency but also enhanced customer satisfaction, with errands resolved in less than 2 minutes compared to 11 minutes previously.

Embracing Strategic Pivots

In the dynamic landscape of SaaS and AI, the ability to pivot strategically is crucial. Many successful companies have undergone significant pivots in their business models to adapt to market changes and achieve growth. For instance, Twitter originated as a podcast service called Odeo before pivoting to a microblogging platform. Similarly, Shopify transitioned from an online snowboarding equipment store to a comprehensive e-commerce platform. Flickr began as an online multiplayer game before becoming a photo-sharing site. Pinterest started as a mobile shopping app named Tote before evolving into a visual discovery platform. These pivots often involve redefining core business assumptions and engaging new resources, technologies, and leadership. Such examples underscore the importance of flexibility and responsiveness in business strategy, especially in the face of technological advancements like agentic AI.

Pivots for SAAS include

1. Build Embedded Agents: Integrate agentic UX within your product. Employ intent-based UI, context memory, and internal Retrieval-Augmented Generation (RAG).
2. Attack via Vertical Orchestration: Control agents across specific vertical domains (e.g., construction, legal, compliance). Examples include Procore, ServiceNow, and Toast.
3. Own the Model Logic: While not necessarily owning the LLM itself, manage your RAG, fine-tuning, and abstraction layers. Utilize tools like CoreWeave, Mistral, and LangGraph for efficient development.
4. Best Ecosystems to Build I: key ecosystems for development include infrastructure providers like CoreWeave, Lambda, and RunPod; model developers such as Mistral and LLaMA 3; orchestration tools like AutoGen, CrewAI, and LangGraph; SaaS innovators including Notion, Intercom, and Deepop; and VC/PE firms like EQT, Point Nine, and Index.

Are you ready to embrace?

About the Author

Jacques Bughin is the CEO of MachaonAdvisory and a former professor of Management. He retired from McKinsey as a senior partner and director of the McKinsey Global Institute. He advises Antler and Fortino Capital, two major VC /PE firms, and serves on the board of several companies.