TechEx North America: AI Confronts the "Project Graveyard"

AI and the "Project Graveyard" Challenge at TechEx North America

On the second day of TechEx North America, the program dedicated to artificial intelligence (AI) and Big Data highlighted a striking expression: the "AI graveyard." This term refers to the large number of AI pilot projects that fail to become sustainable systems. This imagery served as a common thread throughout discussions, focusing on the need to prove the viability of AI projects.

Challenges of Implementing AI in Business

A key session addressed the implementation of AI in businesses, concentrating on return on investment (ROI) and adoption. Discussions illuminated the obstacles faced by pilot projects, the commercial impact of agentic AI, and the transition from experimentation to tangible results. Speakers emphasized the importance of deciding between buying or building solutions, while insisting on the necessity for sustainable ROI and autonomous decision-making. For a system to be successful, it must be effectively adopted, governed, and measured.

Understanding AI Project Failures

The session dedicated to the "AI graveyard" allowed for the identification of recurring failure patterns. Many companies have the financial resources and executive attention to initiate AI projects. However, few succeed in maintaining these projects due to insufficient data quality, poorly designed processes, lack of operational authority, and inadequate risk management.

From Co-Pilot AI to Agentic AI

Another session explored the transition from AI co-pilots to more autonomous AI agents, emphasizing commercial impact over technological novelty. Co-pilots, while useful for individual productivity, have a value that is difficult to quantify. Agentic AI, on the other hand, promises closer integration with business processes but requires clear boundaries. Assessing the quality of these agents' actions is crucial for their success.

Trust and Governance: Crucial Issues

The theme of trust as a competitive advantage was central in the track on the future of AI. Discussions focused on transparency, governance, regulation, and risks, particularly in the banking sector. Elements from Hex were integrated, highlighting the importance of evaluating and governing data agents. For agentic AI to thrive in businesses, formal assessment is essential.

Various Forms of Governance

Governance was addressed from multiple angles: cross-functional, reflecting the diversity of risks associated with AI, and at the data level, where trust relies on quality and traceability. The governance of agent personas and risk stacks was also discussed, underscoring the need for companies to understand the capabilities and limitations of AI agents. In the banking sector, this governance is crucial, as financial services cannot afford vague assurances regarding automation.

Pressure on Digital Transformation

Digital Transformation Week continued the pressure from the second day towards commercial realization. The program focused on real use cases, commercial impact, ROI, and API-based AI agents. Change readiness was a major topic, emphasizing that AI fails when staff routines do not change, managerial incentives remain unchanged, or necessary data is not available at the right time.

Integrating AI into Government Services

Sessions involving the DMV and the City of San José demonstrated how AI and digital transformation can be integrated into government services. In this context, quality is measured by reliability, accessibility, explainability, and public trust. Meanwhile, Dow illustrated how to convert data into financial value, emphasizing that value lies in the connection between working with data and responsible outcomes.

Data Security and Governance in the Cloud

The second day's program at the Cyber Security and Cloud Expo broadened the debate on risks, focusing on AI threats, cloud security, and the "GenAI speed gap." AI is seen as a force transforming both attack and defense, automating defensive work while accelerating abuses and increasing pressure on existing controls.

The Speed Gap and the Zero Trust Response

The term "speed gap" was recurrent, highlighting that business units are adopting generative AI faster than security teams can oversee it. Sessions on jailbreaking and data leaks illustrated this point. If sensitive materials are placed in unapproved tools, or if approved AI systems are poorly delineated, cloud security and data governance become inseparable.

Zero trust was proposed as a solution, with a more robust interpretation that now includes AI systems, agents, and data. Identity is no longer limited to human users; services, agents, and automated workflows also require permission models. Thus, the cloud-first enterprise becomes an environment where identity, data classification, AI governance, and threat detection are integrated into a single control framework.