AI Landscape, Part V: Your Data, Your Process 

The Foundation AI Will Not Fix for You 

Bottom line: AI will not fix a broken process. It will automate that process faster, scale it farther, and make the output look more credible than it may deserve. If the workflow is undocumented or the data foundation is weak, the system will not quietly compensate. It will amplify the weakness and do it behind an interface that can make bad output look finished. 

Supporting context 

This is the part of AI evaluation most organizations want to skip. Data quality and process documentation are not exciting. Nobody gets much energy from saying, “Let’s audit the workflow,” especially when the alternative sounds like, “Let’s deploy an AI agent.” But this is where AI strategy meets operating reality. It is also where many AI efforts either become useful or become expensive theater. 

Three separate but tightly linked questions live here: 

1. Is your data actually ready? Is it accurate, accessible, and structured well enough for an AI system to work against it reliably? 

2. Is your underlying process actually well-defined, or does “the way we do this” only exist in one person’s head, adjusted case by case with no documented log? 

3. What documentation of process and/or standardization of data is already in place? 

These issues compound. Messy data flowing through an undocumented process does not average out to “medium risk.” It creates a system nobody can debug because nobody can point to what “right” was supposed to look like in the first place. If nothing is documented, there is no reliable fallback to the company standard, the approved method, or the judgment pattern the process depends on. 

Data or evidence 

The data supports the pattern. Informatica’s 2025 CDO Insights research found that 43% of data leaders cited data quality, completeness, and readiness as obstacles preventing generative AI pilots from reaching production, while 67% had not successfully moved even half of their pilots into production. Gartner separately reported that 63% of organizations either lacked, or were unsure they had, the data management practices needed for AI; Gartner also predicts that through 2026, organizations will abandon 60% of AI projects that are not supported by AI-ready data. 

The business-value gap is just as clear. MIT Project NANDA’s 2025 report found that despite $30–40 billion in enterprise generative AI investment, 95% of organizations in its study were seeing zero measurable return, with only 5% of integrated pilots extracting material value. McKinsey’s 2025 Global AI Survey tells a similar story from a different angle: 88% of respondents said their organizations use AI in at least one business function, yet only 39% reported enterprise-level EBIT impact. Adoption is widespread. Scaled value is not. 

The signal is not that AI is useless. The signal is that AI value depends on the operating environment around it. McKinsey found that high-performing organizations are more likely to redesign workflows as part of their AI work, not simply layer tools onto existing habits. MIT Project NANDA similarly attributed stalled pilots to brittle workflows, limited contextual learning, and misalignment with day-to-day operations. 

In practical terms, AI is still a tool that depends on human judgment, governance, and accountability. The more consequential the workflow, the more important it is to know what data the system is using, what process it is following, who owns the decision, and how the result will be validated. 

Implications 

Run the data audit before the pilot. Not after. Before. Where does the relevant data live? Who owns it? Which source is trusted? Is it complete enough that you would ask a person to decide from it? If the honest answer is, “It is scattered across four systems and nobody knows which version is current,” then the first project is not AI deployment. The first project is data readiness. 

Document the process before you automate it. If you cannot write down how the task gets done, including the exceptions and judgment calls, you do not yet have a process to automate. You have a habit held together by experienced people who have internalized rules nobody else can see. Automating that does not scale expertise. It locks inconsistency into a system that can now repeat it faster. 

Start smaller than your ambition. Small, well-defined workflows with clean data and clear ownership beat broad, fuzzy initiatives with heroic expectations. The goal is not to prove that AI can do something impressive in a demo. The goal is to prove that human teams and AI systems can work together in a repeatable operating model that improves quality, speed, or focus without weakening accountability. 

Net: When the AI gets it wrong, check the data and the process before blaming the model. In many cases, the defect was already present upstream. The model did not create the problem. It exposed it, packaged it, and delivered it with confidence. 

Where this matters 

This may be the least glamorous article in the series, but it may be the most important. Model selection, ecosystem fit, and security architecture matter. But data quality and process clarity are the foundation those choices depend on every day. Get the foundation right and the technology decisions from earlier in this series become easier to execute well. Get it wrong and no model, no platform, and no amount of security hardening will save the outcome. 

If you are evaluating where AI belongs in your business, start here: identify the workflows where better data, clearer process, and human accountability would create measurable value. That is where LumenForge Advisors helps leaders move from AI interest to AI operating discipline, with practical adoption grounded in business reality, responsible governance, and human-centered execution. 

Sources 

1. Informatica. “Global Data Leaders Seek to Harness the Power of GenAI for AI-Driven Success.” January 28, 2025. https://www.informatica.com/about-us/news/news-releases/2025/01/20250128-global-data-leaders-seek-to-harness-the-power-of-genai-for-ai-driven-success.html 

2. Informatica. CDO Insights 2025: Racing Ahead on GenAI and Data Investments While Navigating Potential Speed Bumps. 2025. https://www.cvisionintl.com/media/ezaj1fsb/informatica-cdo-insights-2025.pdf 

3. Gartner. “Lack of AI-Ready Data Puts AI Projects at Risk.” February 26, 2025. https://www.gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk 

4. MIT Project NANDA. The GenAI Divide: State of AI in Business 2025. July 2025. https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf 

5. McKinsey & Company. “The State of AI in 2025: Agents, Innovation, and Transformation.” November 5, 2025. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai