Skip to main content

> Stack

The Enterprise AI Stack

Incidents where enterprise data complexity breaks the illusion of seamless AI integration.

"The model was state-of-the-art. The enterprise data was state-of-the-90s."

What this stack means

This stack tracks the friction of deploying AI within large organizations fraught with legacy data and strict permissions.

Why this stack exists

Because enterprise data is rarely clean, centralized, or correctly permissioned for AI consumption.

▶ Common Failure Patterns

  • •permission bypass via RAG
  • •garbage-in-garbage-out scaling
  • •context window saturation with boilerplate
  • •AI-washing legacy tools
  • •semantic search over unstructured chaos

Prevention Checklist

  • Enforce data access controls at the retrieval layer.
  • Clean and structure data before feeding it to an LLM.
  • Define clear success metrics for internal AI tools.

Detection Signals

  • The AI confidently summarizing confidential documents for unauthorized users.
  • Employees abandoning the internal AI tool because it only provides generic answers.
  • High API costs driven by inefficient RAG pipelines.

Incidents in The Enterprise AI Stack

Reference
The Procurement Theater StackProcurement and Vendor Theater

SLA More Optimistic Than Reality

"The chaos was predictable."

Pattern: demo-to-contract drift
Read Incident →
Reference
The Security and Governance StackSecurity, Compliance and Audit

Retry Policy Tried Too Hard

"The chaos was predictable."

Pattern: compliance theater
Read Incident →
Reference
The Security and Governance StackSecurity, Compliance and Audit

Architecture Review Became Therapy

"The chaos was predictable."

Pattern: compliance theater
Read Incident →
Reference
The Security and Governance StackSecurity, Compliance and Audit

Release Train Had No Brakes

"The chaos was predictable."

Pattern: compliance theater
Read Incident →
Reference
The Executive Transformation StackAI Transformation Theater

The Incident Commander Needed a Whiteboard

"The chaos was predictable."

Pattern: pilot-without-operating-model
Read Incident →
Reference
The ModelOps StackLLMOps, Evals and Observability

The Whiteboard Lied Beautifully

"The chaos was predictable."

Pattern: confidence without verification
Read Incident →
Reference
The ModelOps StackLLMOps, Evals and Observability

The Model Hallucinated Confidence

"The chaos was predictable."

Pattern: confidence without verification
Read Incident →
Reference
The ModelOps StackLLMOps, Evals and Observability

The Demo Worked in the Recording

"The chaos was predictable."

Pattern: confidence without verification
Read Incident →
Reference
The Procurement Theater StackProcurement and Vendor Theater

The Recording Became the Product

"The chaos was predictable."

Pattern: demo-to-contract drift
Read Incident →
Reference
The ModelOps StackLLMOps, Evals and Observability

The AI Strategy Was a Slide Deck

"The chaos was predictable."

Pattern: confidence without verification
Read Incident →
Reference
The ModelOps StackLLMOps, Evals and Observability

The Slide Deck Asked for a Platform

"The chaos was predictable."

Pattern: confidence without verification
Read Incident →
Reference
The ModelOps StackLLMOps, Evals and Observability

The Platform Asked for Ownership

"The chaos was predictable."

Pattern: confidence without verification
Read Incident →
Reference
EP51The ModelOps StackLLMOps, Evals and Observability

The Model Hallucinated Confidence

"The core technical takeaway from 'The Model Hallucinated Confidence' is that isolated decisions scale poorly."

Pattern: confidence without verification
Read Incident →

The Enterprise AI Stack - Frequently Asked Questions

What is this stack?

AI initiatives trapped in corporate governance.

AI Summary

Incidents where enterprise data complexity breaks the illusion of seamless AI integration.