TOTVS and the question of the moment: will AI replace ERP?

As AI agents become increasingly capable of automating a wide range of tasks, questions arise as to whether SaaS companies such as TOTVS could be replaced.

Luca Damico Terada (Junior Member)

In recent days, shares of software companies, especially Software as a Service (SaaS) companies, have entered a wave of selling that appears more emotional than rational, with sharp declines, elevated volatility, and a market attempting to price in what AI may represent for the sector. In the United States, the software index (IGV) has accumulated a significant decline since the end of October, and the discussion around the "existential threat" of AI has become the kind of narrative that accelerates panic, to the point where some companies have announced billion-dollar buybacks and still failed to calm the market.

Chart, Total software and services market (USD Billion), IGV (%) (Sources: ABES, Brazil Journal)

OpenAI and Claude helped intensify this discussion because, in recent weeks, they moved beyond talking only about better models and began selling a much more direct idea of AI as execution, not just response. OpenAI launched Frontier, a platform for companies to create, deploy, and manage agents with shared context and clear permissions, that is, agents that, beyond a chat interface, connect to tools and operate processes. In parallel, Anthropic (Claude) announced a suite of corporate plug-ins designed for specific tasks, from investment banking to HR and engineering, connecting Claude to services such as Slack, DocuSign, and data partners. According to Reuters, this type of announcement came shortly after a "trigger moment" that fueled market aversion to the software sector. When two of the leading AI companies begin launching work-ready agents and distributing them through consulting firms and integrations, the investor reading shifts to one where this is no longer just technological innovation, but an attempt to capture the budget that previously went to traditional tools, such as ERP, an enterprise management system that integrates the main areas of a company, such as finance, procurement, and inventory, into a single platform.

Against this backdrop, Brazilian giant TOTVS also suffered from the sector's repricing, falling around 13% in a single day. This occurred because the company's core operations are tied to SaaS solutions, most notably ERPs and management systems used in companies' day-to-day operations. Thus, in a market that began pricing in the risk of AI rendering parts of software less relevant, TOTVS ended up being impacted.

For CEO Dennis Herszkowicz, the idea of the ERP's "death" is nothing but "investor madness." Even so, TOTVS has used this context to respond to the stock's devaluation with capital allocation initiatives, such as share buyback programs. In February 2026, a buyback program of up to 20 million shares was announced, valid through February 2027. Additionally, the company announced the launch of LYNN, TOTVS's AI platform aimed at businesses, created to serve as the foundation for the development and use of AI agents and features within its management solutions, with a focus on security, governance, and corporate applications. With this, the company seeks to incorporate AI into its portfolio to expand its customer base and reduce the risk of losing relevance that the market has begun to price in.

The recent sector correction did not spread evenly across companies, but the market's new lens appears to be the same. With AI becoming more present at the front end, investors have begun looking at cloud recurrence and growth with greater skepticism, even when indicators remain healthy. In Brazil, Senior had been delivering progress in cloud and recurring revenues, but was caught in the same risk package that hit software globally. Abroad, SAP and Oracle continue to show cloud expansion and a resilient customer base, but the narrative has shifted from predictable subscription-based growth to the question of where value will be captured as new AI layers take center stage. This shift is subtle, but has been sufficient to compress multiples. It does not mean that operations changed from one quarter to the next, but rather that the market became less willing to pay a premium for predictability when the dominant discourse shifted to automation, agents, and platforms that promise to shorten the path between intent and execution.

In this shift of lens, it makes sense to separate what is interface from what is infrastructure. The ERP is the transactional backbone that ties together a company's critical processes and data, and it is precisely this function as a "system of record" that makes a direct replacement by AI unlikely. The reason is less technological and more practical, because the ERP carries tax and accounting rules, controls, integrations with dozens of systems, and an audit trail that must be stable, repeatable, and accepted by areas such as finance, compliance, and IT. Replacing this foundation requires historical data migration, process revalidation, and a level of operational risk that few companies are willing to assume, especially when any failure affects billing, taxes, payments, and financial closing. For this reason, the most plausible trend is that AI will capture value as a complementary layer, automating surrounding tasks and decisions, while the ERP remains the "engine" that ensures consistency and governance.

Chart, ERP implementation barriers (Source: ABES)

For TOTVS, the logic of the ERP as a "system of record" helps put the AI discussion in the right place. The immediate risk is not a mass replacement of the transactional layer, which is costly and risky for the client, but rather AI beginning to mediate ERP usage, drawing the user's attention toward copilots, automations, and more direct workflows. As this layer gains ground, the comparison ceases to be solely between ERPs and begins to include who can shorten work cycles and reduce friction in day-to-day routines. This also shifts what drives the market's reading, placing more weight on adoption signals and incremental monetization than on broad promises. In the end, the dispute ceases to be about replacing the foundation and becomes about who emerges as the preferred interface in daily operations, influencing contract expansion, pricing of additional layers, and the perception of recurring revenue trajectory.

In short, everything comes down to the question that has become the center of the discussion, whether AI will replace ERP, and to how TOTVS enters this debate in Brazil. The recent sector repricing happened more due to a narrative shift than an immediate deterioration of fundamentals, with investors trying to understand where value will be captured as AI solutions promise more direct execution. At the same time, the nature of the ERP as a transactional backbone, filled with integrations, rules, and control requirements, makes a rapid replacement of the core system unlikely. The more plausible discussion ceases to be whether "the ERP will end" and becomes about who controls the interaction and automation layer around it, and it is on this point that the discussion around TOTVS tends to converge.