Salesforce’s John Taschek Unveils 12 Rules for Enterprise Agentic AI as Pilot Failures Persist
Updated
Updated · ZDNet · Jun 25
Salesforce’s John Taschek Unveils 12 Rules for Enterprise Agentic AI as Pilot Failures Persist
1 articles · Updated · ZDNet · Jun 25
Summary
John Taschek, Salesforce’s chief market strategy officer, published 12 vendor-neutral rules for enterprise agentic AI, framing them as a benchmark for production deployments rather than pilot demos.
The framework argues most agentic AI failures are architectural, not model failures, with weak data lineage, stale data, missing semantics, poor observability and absent guardrails repeatedly undermining deployments.
Salesforce says its lessons come from more than 20,000 production deployments, where post-launch management dominates the work and trust must be earned through fairness, explainability, consent controls and hallucination prevention.
More than half of US desk workers describe themselves as AI skeptics, while Informatica found over half of agentic AI adopters cite data quality and retrieval as deployment barriers.
The rules push companies from siloed pilots toward systemic AI that measures business outcomes, preserves enterprise control and supports human-agent collaboration as adoption spreads across regulated sectors and government.
As firms embrace agentic AI, why is the crucial governance needed to prevent system failure being dangerously overlooked?
Are enterprises building AI on a house of cards by failing to solve their fundamental data chaos first?
If AI cannot gain experience, how can we prevent it from eroding the expert human knowledge our businesses rely on?
Building Trust in Agentic AI: Applying Taschek’s 12 Rules to Overcome the 95% Failure Rate in Enterprise Pilots
Overview
John Taschek's 12 Rules provide a foundational framework for responsible enterprise agentic AI, guiding organizations to move beyond incremental updates and fundamentally restructure how autonomous AI operates. The rules emphasize a pragmatic, disciplined approach, starting with contained, high-value use cases to stabilize complex process loops and secure data foundations. This step-by-step strategy helps build confidence and control, ensuring that agentic AI is adopted safely and effectively. By embodying a spirit of trust and operational discipline, Taschek’s framework supports enterprises in managing the challenges of agentic transformation and lays the groundwork for scalable, reliable AI integration.