In a recent LinkedIn post, Luís Rodrigues tackles a pervasive misconception surrounding artificial intelligence, arguing that the technology’s failures are often rooted not in the algorithms themselves, but in foundational data issues that leaders are neglecting. Rodrigues, a proponent of practical AI implementation, shared his perspective on why many AI initiatives falter, drawing from his own experiences.
Rodrigues begins by stating a core problem: AI did not create data issues; it merely amplified them. He recounts a past experience at QuickCheck where a meticulously built machine learning model failed for clients post-deployment, not due to algorithmic flaws, but because of poor data quality.
“The model was great but the data was terrible.”
This anecdote sets the stage for his broader critique of how many organizations approach AI. Rodrigues points out that while roadmaps are often filled with advanced concepts like Agents, MCPs, and A2A automations, a fundamental question frequently goes unanswered at the board level: “Who owns the data?” The silence that follows, he suggests, indicates a critical governance gap.
The Illusion of AI Without Data Governance
Rodrigues contends that most companies are not struggling with an inherent AI problem, but rather a data problem they are actively ignoring. He paints a stark picture of the typical data landscape within these organizations:
- Zero governance
- Zero data lineage
- Unmaintained data pipelines
- Data duplication across multiple systems
- Data lakes that have devolved into “data swamps”
He criticizes the common practice of layering advanced AI solutions, such as agents, on top of this unstable foundation, expecting miraculous results. “Then they put an agent on top of it and expect magic,” Rodrigues writes.
“Most companies don’t have an AI problem. They have a data problem that they pretend doesn’t exist.”
“Data-First” as a Leadership Imperative
According to Luís Rodrigues, adopting a “data-first” approach is not merely a technical strategy but a significant leadership decision. It requires assigning clear accountability for data assets, complete with a dedicated name and budget. He emphasizes that the success or failure of AI models is directly tied to the quality and management of the underlying data.
“AI doesn’t fail because of the model. It fails because nobody fixed what’s underneath.”
Rodrigues likens the essential groundwork of data management to a building’s foundation. While foundational work might be less glamorous than implementing cutting-edge AI, it is indispensable for the stability and success of the entire structure. “People don’t like the foundation work, but they’re the only reason the rest doesn’t collapse,” he asserts.
The 5-Second Test for Data Ownership
To underscore the urgency and clarity required, Rodrigues proposes a simple diagnostic: if an organization cannot answer the question of data ownership within five seconds, it signals a deep-seated issue. This “5-second test” serves as a quick litmus test for the health of an organization’s data strategy and governance.
Luís Rodrigues concludes that true AI success hinges on addressing these fundamental data challenges head-on, driven by decisive leadership that prioritizes and invests in robust data governance and ownership. He invites readers to subscribe to his newsletter, “Build What Matters,” for further insights into AI agents and emerging workflows.
📝 About This Content
This article is based on insights shared by Luís Rodrigues on LinkedIn.
📅 Originally posted on October 3, 2026 | View original post on LinkedIn →