The Method.
Alchemists were the first to keep laboratory journals — to write down what they mixed, what they burned, and what failed. The metaphor is old; the discipline is current. Every engagement runs through five operations. Each has a defined output. Nothing rests on inspiration alone.
Calcinatio — Define the problem
Burn away the assumptions. We establish what is actually known, what only seems known, and where the genuine uncertainty lies: existing methods, the measurable baseline, the criteria that will count as success.
A research question worth answering.
Solutio — Design the research
Dissolve the problem in its data. Methodology, literature and technology review, dataset exploration, data-quality assessment, hypotheses and their alternatives, and a plan of experiments designed to distinguish them.
Methodology + experimental plan.
Separatio — Experiment
Separate what works from what doesn't. Algorithms are built and trained, architectures compared, features engineered, results benchmarked — iteratively, against the baseline, with every run recorded.
Experimental models, algorithms, documented results.
Coagulatio — Condense into a prototype
Thicken knowledge into substance. The selected approach becomes a proof of concept: experimental software, a pilot system, integration into a test environment — prepared for controlled real-world trial.
A working prototype / pilot system.
Distillatio — Validate
Purify the result. Testing against the baseline and the success criteria, robustness checks, analysis of what held and what didn't, technical documentation, identification of the intellectual property created.
A validated deliverable + a research report.
Stage III sometimes proves a hypothesis wrong. That, too, is a research result — documented, explained, and considerably cheaper than discovering the same thing after a full-scale implementation. We report it exactly as we would report success.