Evolution
Learning, adapting, and compounding capability over time. Evolution methods turn work into a measurable improvement loop, gate change behind evaluation, and ensure both the system and the methodology itself improve through deliberate, evidence-driven iteration.
8 methods| ID | Method | Description |
|---|---|---|
| FG-0801 | Operational Flywheel 3 | Work output feeds back into system improvement. Every task, challenge, or engagement produces artifacts that can be captured, structured, and used to make the system better at the next task. This is a data flywheel: solve problems, capture each solution as structured data, use that data to improve tooling or fine-tune the model, let the improved system take on harder problems, and repeat. It is a concrete pipeline for continual improvement in which today's work becomes tomorrow's training signal. |
| FG-0802 | Comparison-Driven Improvement 3 | A deliberate loop: solve a challenge, write it up, compare your writeup against reference material, extract delta training pairs from the comparison, feed those pairs into model training, and solve the next challenge better. The deltas are a form of active learning: the disagreements between your approach and the reference are the most informative examples, and they form candidate preference pairs for methods like DPO. Each revolution produces both data and capability. The comparison step is where the real value lives: your approach versus the reference reveals methodology gaps, alternative techniques, and efficiency differences that pure self-review misses. |
| FG-0803 | Eval-Gated Promotion 3 | No new version reaches production on intuition. Eval-Gated Promotion requires every candidate model, agent, prompt, or tool to pass a defined evaluation suite before it replaces the incumbent. The suite is a fixed set of prompts or tasks, scored on the dimensions that matter, run identically against the current production version and each challenger. The version that wins on the metrics advances; the rest are documented with their specific failure modes. This turns version selection into a measured comparison rather than a judgment call. |
| FG-0804 | Regression Guardrails 3 | A new version can improve on one axis while silently degrading another. Regression Guardrails are a frozen, held-out set of checks that a change must continue to pass before it ships. Where Eval-Gated Promotion asks whether a candidate is better, the guardrail asks a separate question: did anything that used to work just break? The guardrail set is preserved unchanged across versions precisely so that it can detect when a capability you already had has regressed. |
| FG-0805 | Feedback Capture Loop 3 | Improvement needs a signal, and the signal is feedback. The Feedback Capture Loop systematically records the human and automated judgments produced during normal work: corrections, approvals and rejections, preference between two outputs, and pass or fail against a verifier. Captured in a consistent format, these judgments become the training signal that drives the flywheels. These are the same kinds of signal that underlie supervised, preference-based, and reinforcement-style model training, gathered through daily practice rather than in a one-time labeling project. |
| FG-0806 | Corpus Curation 3 | The improvement corpus is the accumulated body of structured data the system learns from. Left to grow on its own, it collects duplicates, contradictions, and low-value entries that degrade the signal faster than volume improves it. Corpus Curation treats the corpus as a maintained asset: deduplicated, quality-gated at intake, versioned, and pruned. The quality of what the system learns is bounded by the quality of the corpus, so curation is the difference between a flywheel that compounds and one that drifts. |
| FG-0807 | Capability Versioning 3 | Models, agents, and prompts change over time, and an undocumented change is an untraceable one. Capability Versioning applies the discipline of software versioning to the components that produce behavior: every model, agent profile, and significant prompt is versioned, reproducible from its inputs, and reversible to a known-good prior state. When behavior changes, versioning is what lets you say exactly what changed, compare it against what came before, and return to the previous version if the new one regresses. |
| FG-0808 | Methodology Evolution 3 | The framework that governs the work must itself improve, or it ossifies into bureaucracy. Methodology Evolution treats the methodology as a versioned, evidence-driven artifact: methods are added when a pattern proves out across repeated practice, refined as evidence accumulates, and deprecated when superseded. Most methods exist because something went wrong and the system was built to prevent it, which means the methodology is never finished. It evolves on evidence drawn from the daily work it governs. |