Goldratt applied to AI adoption
AI driven modernization in large enterprises typically starts with an urgent focus on AI tech stack and hiring the best AI talent. Ironically these transformation initiatives can only go at the speed the slowest link in the chain permits. This link is typically systems that GenAI bots and AI agents depend on - systems that were built for humans - not AI agents.
Goldratt's Theory of Constraints focuses on exploiting the weakest link first. Focusing on constraints rather than strengths will usually seem counterintuitive as it does to the protagonist of the book. Yet, it makes sense once you take in the whole picture rather than just the 'AI development and deployment' components in the transformation journey.
Understanding the Theory of Constraints
The Theory of Constraints (TOC) is a management philosophy developed by Eliyahu M. Goldratt, focusing on identifying and eliminating constraints to improve overall system performance. It emphasizes that every system has at least one constraint that limits its ability to achieve its goals.
Key principles of Theory of Constraints
The core principles behind TOC are:- Convergence Principle: A complex system is simpler to manage because any adjustment or correction to one part of the system affects the entire system as a whole. Improving one part (usually the constraint) improves the whole system.
- Consistency Principle: Any internal conflict or issue within a system results from at least one flawed assumption. Identifying and resolving these flawed assumptions is key to improving the system.
- Respect Principle: The principle of respect acknowledges that people are inherently good and deserve respect even when mistakes occur, allowing flexibility in management involving human factors.
ToC also outlines five basic steps to manage constraints.
- Identify the constraint.
- Exploit the constraint.
- Subordinate everything else to the constraint (align the whole system to support the constraint).
- Elevate the constraint.
- Avoid inertia by repeating the process continuously to find and manage new constraints.
Applying the Theory of Constraints to AI transformation
Most AI platforms and tools make it look very easy to pivot to an AI driven workplace. Accessing volumes of data through GenAI chatbots, building prediction into apps, creating AI agents and running an agentic AI workflow are typically on the radar of most enterprise companies today. Let's look at the various parts of this puzzle.The AI journey for most large enterprises involves:
- Creating awareness about and momentum for the use of AI in business, engineering and operations.
- Identifying use cases
- Building and implementing AI agents
- Creating a feedback loop about AI features, utilization and impact through user and application telemetry.
Within this list the third step can be expanded into a few component elements.
- AI stack
- AI engineers
- MLOps