TechEx North America: The Unseen Pitfalls of Enterprise AI
Day two at TechEx North America revealed the stark realities of enterprise AI adoption. Amidst the discussions, trust and governance emerged as the unsung heroes.
TechEx North America's second day put the spotlight on a grim reality: the 'AI graveyard' where pilots go to die. This term summed up the trials of transforming flashy AI experiments into durable systems. The question that echoed throughout was simple: where's the proof?
From Pilots to Practice
The sessions focused on the middle ground of AI work that companies often stumble upon, transitioning from pilot projects to impactful systems. Talks covered the usual suspects like stalled pilots and agentic AI, which promises business impact but often fails to deliver. The theme was clear: AI needs governance, adoption, and measurement to succeed.
The session on 'AI graveyard' was particularly enlightening. Many enterprises have funds to kickstart AI projects and executives eager to showcase them. But they lack the data quality, process design, and risk management to keep these projects alive. It's a pattern we see too often, and it's sabotaging progress.
Beyond Copilots: Agentic AI and Business Impact
Another highlight was the shift from AI copilots to agentic AI, emphasizing business impact over novelty. While copilots boost individual productivity, their value often remains elusive. Agents, though promising, demand stricter boundaries. If an agent acts within systems, its actions must be judged by their quality.
This discussion tied into the Future of AI track, where trust was hailed as a strategic advantage. Speed may be enticing, but trust, transparency, and governance emerged as critical elements. Without formal evaluation, agentic AI won't mature in enterprise settings.
Governance: The Backbone of AI Success
Governance was a recurring theme. Cross-functional governance highlighted that AI risk isn't just a concern for legal or security teams. Data layer governance was about ensuring trust through data lineage and quality. It's clear: understanding what an AI agent can and can't do is important, especially in sectors like banking where automation leaves no room for ambiguity.
Sessions on digital transformation underscored these themes with real-world use cases. Change readiness, government service transformation, and data-driven financial value were turning point discussions. AI often fails because people resist change or important data remains inaccessible. The City of San Jose and the DMV exemplified how AI can transform government services by prioritizing reliability, access, and public trust.
Cyber Security: Bridging the Velocity Gap
The Cyber Security and Cloud Expo expanded on risk, with AI-led threats and cloud security taking center stage. The 'GenAI velocity gap' was a buzzword, highlighting how fast businesses adopt generative AI compared to their security teams' ability to manage it. If sensitive data lands in unsanctioned tools, cloud security and data governance become intertwined issues.
Zero trust was proposed as a solution, but with a twist: it must now encompass AI systems and agents. Identity isn't just about human users anymore. Services, agents, and workflows also require stringent permission models. The cloud-first enterprise now integrates identity, data classification, AI governance, and threat detection into a unified control mechanism.
Get AI news in your inbox
Daily digest of what matters in AI.
Key Terms Explained
Agentic AI refers to AI systems that can autonomously plan, execute multi-step tasks, use tools, and make decisions with minimal human oversight.
An autonomous AI system that can perceive its environment, make decisions, and take actions to achieve goals.
A machine learning task where the model assigns input data to predefined categories.
The process of measuring how well an AI model performs on its intended task.