O “SaaSocalypse” é, na verdade, uma Crise de Arquitetura e Identidade

Executive Summary

Gartner’s warning projecting $234 billion in software licenses at risk by 2030 (the “SaaSocalypse”) has exposed the collapse of the pay-per-seat model. As discussed in a recent analysis by *Exame*, the thesis that “SaaS is dying” masks a deeper infrastructural transformation: value has shifted from the graphical interface to autonomous execution. This article analyzes this transition and outlines how enterprise engineering must evolve from screen-building (UI-centric) to the orchestration of autonomy, context-based security (“On-Behalf-Of”), and experience decoupling.

Core Thesis

Traditional SaaS is not on the verge of extinction due to functional failure, but rather due to interface inefficiency. The per-seat model was designed for the era of human navigation. When AI agents execute operational workflows in milliseconds via API calls, the “logged-in users” metric loses its reason for existing. The market will shift from interface to execution, consolidating “System of Record” SaaS—where software maintains transactional integrity and identity governance (RBAC/OAuth2)—while agentic intelligence takes over execution.

+-----------------------------------------------------------------------------------+
|                            TRADICIONAL ERA  (SaaS GUI)                             |
|  [Human User] -----> [Graphic Interface (GUI)] -----> [SaaS / DB]             |
|  Metric: Pay-Per-Seat                                     |
+-----------------------------------------------------------------------------------+
                                          |
                                          v
+-----------------------------------------------------------------------------------+
|                          AGENTIC ERA (Headless Core)                             |
|  [Human] -> [Chat/Voice] -> [AI Agent] -> [API Gateway] -> [Core System / DB]   |
|  Metric: Transaction, Outcomes & API consume (On-Behalf-Of RBAC)               |
+-----------------------------------------------------------------------------------+

Key Insights

The Nuance of “On-Behalf-Of”: The per-seat model will not disappear overnight. Due to regulatory and governance requirements (LGPD, SOX, Bacen), agents must access legacy systems using the credentials and tokens of the user they represent (the “On-Behalf-Of” flow), thereby extending the lifespan of licenses focused on access control at the source.
Emergence of New Pricing Models: Monetization will shift toward Headless/Identity (API-only), Outcome-Based (payment per completed transaction), Consumption-Based (computational volume), and Value-Share (a fraction of the efficiency generated) models.
Architectural Revolution Based on 4 Pillars: A transition from a UI-centric paradigm to an API-based and Event-First infrastructure (via the MCP protocol), a Data Foundation (real-time RAG), Decoupled Agentic UX (Invisible Apps within WhatsApp/Teams), and a Governance Layer (Zero Trust and FinOps).

1. The End of the “Pay-Per-Seat” Era and the Interface Transition

As highlighted in the *Exame* article, the pay-per-seat model underpinned vendor valuations by linking revenue to operational headcount. When AI agents replace manual data entry with direct API interactions, “wrapper” software (interfaces built over simple data entry forms) loses its financial justification. Basic helpdesk tools, passive report generators, and unintelligent connectors are becoming obsolete. However, this transition requires maintaining rigorous corporate security standards. If an agent accesses an ERP system using a generic administrator credential, an auditability blind spot is created. The architecture demands identity propagation at the source (the “On-Behalf-Of” flow): the request to the backend system injects the human user’s token, forcing the system-of-record database to enforce its own Role-Based Access Control (RBAC). This imperative solidifies “Headless” or “Identity” licenses, under which companies pay a fraction of the traditional cost to maintain active access governance, even without launching the graphical interface.

2. Agentic Architecture and Legacy Connectivity

Modern software development restructures engineering around four crucial pillars: an API & Event-First foundation adopting MCP to standardize tool calls; a Data Foundation featuring real-time pipelines and advanced RAG; a Decoupled Experience (“Invisible App”) extending user journeys to channels like WhatsApp and Teams; and Governance & Control Plane mechanisms for cost control (FinOps) and guardrails.

               +------------------------------------+
                |   Channels (WhatsApp, Teams, Voice)    |
                +------------------------------------+
                                  |
                                  v
                +------------------------------------+
                |  AI Agent Orchestrator     |
                +------------------------------------+
                                  |
        (User Token Propagation / On-Behalf-Of Flow)
                                  |
                                  v
                +------------------------------------+
                |   API Gateway & MCP Servers        |
                |   (Source-based RBAC Validation)   |
                +------------------------------------+
                                  |
        +-------------------------+-------------------------+
        |                                                   |
 [Access Granted]                                   [Access Denied]
 (Access authorized in Core)                           (403 Forbidden)

Replacing Systems of Record (ERPs, Mainframes, Core Banking) is financially unfeasible and, for the time being, unnecessary. Connectivity with legacy systems is achieved via a Three-Layer Security Bridge: a conversational front-end with user identity authentication; API Gateways/Wrappers to translate modern calls (OAuth2/JWT); and Source-based RBAC propagation without duplicating rules within the AI.

3. Executive Guidelines for IT Leadership

For the technology executive, the transition to agentic systems is not merely an experimental challenge, but a decision regarding capital governance and architecture. The role of leadership is to move beyond “innovation theater” and guide the organization with pragmatism: True AI leadership requires auditing the portfolio’s TCO to replace paid, per-seat licenses—which serve merely as data inputs—with autonomous execution workflows. Instead of rewriting the company’s core systems, focus investments on exposing APIs and MCP servers atop legacy systems of record, thereby ensuring access governance. Finally, evolve the company’s operational culture from micromanagement (“human-in-the-loop”) to exception supervision and value auditing (“human-on-the-loop”), channeling the freed-up capacity exclusively into initiatives that impact financial margins and the corporate balance sheet.

A bout the Author

Rodrigo Bornholdt is the Co-founder and Chief Technology Officer of Zappts. He specializes in Software Architecture and Artificial Intelligence, possessing extensive experience in leading technology teams, developing complex systems, and driving innovation within business strategies.

About Zappts

With 12 years of experience, Zappts is a technology and innovation company recognized as a leader in Agentic Transformation for large corporations. The company has successfully delivered over 280 projects and logged more than 1 million engineering hours across sectors such as finance, healthcare, retail, and energy. It is the creator of the “Panorama da IA no Brasil” (AI Landscape in Brazil)—a study mapping the country’s technological maturity—and a benchmark in implementing AI agents integrated into core business operations, with a focus on governance, ROI, and operational efficiency. Click here to know more.