Just as falling leaves follow chaotic paths yet always reach the ground, ABCD's AI agents explore wildly different approaches while being pulled toward your requirements like a strange attractor. Unpredictable creativity, predictable results.
Attractor-Based Convergent Development (ABCD) harnesses this chaos theory principle for software development: multiple AI agents independently explore different solutions while mathematical forces ensure they all converge on working code.
This website demonstrates the process in action—every line was generated by an AI agent working within ABCD's framework, with complete source code available for study.
The "A" in ABCD, the attractor is the fundamental force that drives agents towards the spec.
This is literally the "strange attractor" - without it, you have chaos without convergence. The tests are what create the gravitational pull toward viable solutions.
In ABCD, the human software engineer(s) are responsible for providing sensible project scaffolding to the agents to avoid development iterations on basic needs.
Without scaffolding, agents would waste enormous time and compute resources solving already-solved problems like "how to set up a web server" instead of focusing on the unique solution space.
Guardrails prevent the generation of unsafe or malicious code, whether intentional or not.
AI agents can and will generate unsafe code if not constrained. Security isn't optional when you're letting AI systems write production software.
Control what external dependencies agents can use by defining resources through a whitelist or blacklist approach.
Approved resources prevent agents from pulling in problematic dependencies, ensures compliance with organizational standards, and maintains architectural consistency.