Blab Tech
Legacy ERP Vendors Face AI-Driven Shift as SaaSpocalypse Looms
On June 18, 2026, Younglimwon Soft Lab’s executive vice president Ho Woong‑ki delivered a stark warning at the company’s EBSC 2026 conference in Seoul: traditional enterprise resource planning (ERP) vendors must act fast or risk becoming obsolete. Ho said that the rise of autonomous AI agents is eroding the value of software built around human‑oriented interfaces.
Ho’s remarks come amid a broader industry debate about the so‑called “SaaSpocalypse.” The term describes a perceived threat to software‑as‑a‑service (SaaS) companies as large language models and agentic AI systems gain the ability to perform tasks that previously required human operators. A Gartner report cited at the conference projects that 40 % of enterprise applications will embed AI agents by 2026.
At the heart of Ho’s argument is the idea that AI agents can retrieve data, analyze information, and execute business processes without a user interface. He compared the shift to a “dark factory,” where manufacturing runs fully automated, and called the next evolution of software “dark software.” In such an environment, software providers survive only if they are chosen by AI agents.
Legacy ERP platforms, such as those offered by Younglimwon, traditionally expose functionality through web portals or desktop applications. Ho noted that AI agents require application programming interfaces (APIs) that are optimized for machine consumption rather than human interaction. He added that the company is exploring ways to design APIs that enable AI agents to make better decisions when interacting with enterprise systems.
Data quality and causal understanding are also critical, Ho said. “Having high‑quality data is important, but utilizing that data is even more critical,” he told the audience. “If you simply feed data into AI systems, they often fail to interpret it properly.” He highlighted the need for ontology‑based frameworks that encode business context, relationships, and causality so that AI can learn more effectively from enterprise data.
Another trend Ho highlighted is the growing importance of forward‑deployed engineers (FDEs). Popularized by companies such as Palantir, the FDE model places engineers directly in customer environments to gather feedback and understand workflows firsthand. “They are not sitting in offices making assumptions,” Ho said. “They develop solutions based on real‑world experiences and operational feedback.” Younglimwon has adopted a similar approach, helping customers convert enterprise data into AI‑friendly formats.
The impact on business value is complex. Ho cautioned that efficiency gains do not automatically translate into productivity improvements. “If a task that previously required three hours can now be completed in one hour using AI, efficiency has improved, but productivity has not necessarily increased,” he explained. He urged organizations to redesign job responsibilities, performance evaluation systems, and compensation structures to capture the full benefits of AI adoption.
Pricing models are also expected to shift. Traditional seat‑based licensing, which charges based on employee counts, may give way to usage‑based structures better suited for AI agent environments. Ho noted that global ERP providers such as SAP and Salesforce are already moving aggressively in this direction.
The narrative of a SaaSpocalypse has already affected market sentiment. Some software companies have seen sharp declines in share prices following concerns about the long‑term viability of their business models. Ho acknowledged that historical backend functions and core integration capabilities will continue to exist, but emphasized that software companies need to build competitiveness around those capabilities.
In short, the rise of autonomous AI agents is forcing legacy ERP vendors to rethink architecture, data strategy, API design, and pricing. Companies that can align their products with the needs of AI agents—by providing machine‑readable APIs, high‑quality data, and real‑world engineering support—will be better positioned to survive the transition.
The next few months will likely see increased experimentation with agent‑enabled workflows, further shifts toward usage‑based pricing, and a broader industry conversation about how to measure the true value of AI in enterprise software. Until then, legacy ERP vendors face a clear choice: adapt or risk becoming obsolete in an AI‑driven ecosystem.
Ho’s remarks come amid a broader industry debate about the so‑called “SaaSpocalypse.” The term describes a perceived threat to software‑as‑a‑service (SaaS) companies as large language models and agentic AI systems gain the ability to perform tasks that previously required human operators. A Gartner report cited at the conference projects that 40 % of enterprise applications will embed AI agents by 2026.
At the heart of Ho’s argument is the idea that AI agents can retrieve data, analyze information, and execute business processes without a user interface. He compared the shift to a “dark factory,” where manufacturing runs fully automated, and called the next evolution of software “dark software.” In such an environment, software providers survive only if they are chosen by AI agents.
Legacy ERP platforms, such as those offered by Younglimwon, traditionally expose functionality through web portals or desktop applications. Ho noted that AI agents require application programming interfaces (APIs) that are optimized for machine consumption rather than human interaction. He added that the company is exploring ways to design APIs that enable AI agents to make better decisions when interacting with enterprise systems.
Data quality and causal understanding are also critical, Ho said. “Having high‑quality data is important, but utilizing that data is even more critical,” he told the audience. “If you simply feed data into AI systems, they often fail to interpret it properly.” He highlighted the need for ontology‑based frameworks that encode business context, relationships, and causality so that AI can learn more effectively from enterprise data.
Another trend Ho highlighted is the growing importance of forward‑deployed engineers (FDEs). Popularized by companies such as Palantir, the FDE model places engineers directly in customer environments to gather feedback and understand workflows firsthand. “They are not sitting in offices making assumptions,” Ho said. “They develop solutions based on real‑world experiences and operational feedback.” Younglimwon has adopted a similar approach, helping customers convert enterprise data into AI‑friendly formats.
The impact on business value is complex. Ho cautioned that efficiency gains do not automatically translate into productivity improvements. “If a task that previously required three hours can now be completed in one hour using AI, efficiency has improved, but productivity has not necessarily increased,” he explained. He urged organizations to redesign job responsibilities, performance evaluation systems, and compensation structures to capture the full benefits of AI adoption.
Pricing models are also expected to shift. Traditional seat‑based licensing, which charges based on employee counts, may give way to usage‑based structures better suited for AI agent environments. Ho noted that global ERP providers such as SAP and Salesforce are already moving aggressively in this direction.
The narrative of a SaaSpocalypse has already affected market sentiment. Some software companies have seen sharp declines in share prices following concerns about the long‑term viability of their business models. Ho acknowledged that historical backend functions and core integration capabilities will continue to exist, but emphasized that software companies need to build competitiveness around those capabilities.
In short, the rise of autonomous AI agents is forcing legacy ERP vendors to rethink architecture, data strategy, API design, and pricing. Companies that can align their products with the needs of AI agents—by providing machine‑readable APIs, high‑quality data, and real‑world engineering support—will be better positioned to survive the transition.
The next few months will likely see increased experimentation with agent‑enabled workflows, further shifts toward usage‑based pricing, and a broader industry conversation about how to measure the true value of AI in enterprise software. Until then, legacy ERP vendors face a clear choice: adapt or risk becoming obsolete in an AI‑driven ecosystem.
