Comparisons · 2026
Four reconciled decisions — AI-assisted vs agentic, vibe coding's three faces. Each comparison is scored on criteria that matter, with the verdict named by use case. No affiliate links, no star ratings, no 'winner' chest-thumping.
Ledger contents
Each row is a closed account — criteria audited, verdict named by use case. Open the one that maps to the call you’re trying to make.
Most recently balanced
Verdict · audited 2026-04-20 · 5 criteria · balance 5 levels
Five levels. Level 4 for serious codebases; don't skip past Level 3 without reason.
How I score
Criteria are picked per comparison — whatever the decision hinges on, not a generic feature matrix. A tie means the row doesn't separate the options on that axis; the balance number reads wins-losses-ties (so 5–2·1 = five rows for the right column, two for the left, one tie). Every verdict names a use case, not a champion — AI-assisted for throwaway scripts, agentic for serious codebases. I reconcile a ledger when I publish it; I re-balance it when I ship enough work to change a row.
Adjacent shelves
Get the next close-out
Usually one a month. Each one is a reconciled comparison with a verdict and criteria — no weekly digest, no round-up, no promotion.