AI Completion Asymptote
AI-assisted work reliably reaches "good" (~80%) fast, but the final stretch to "great" (the last 20%) flattens into an asymptote that more prompting often cannot cross — and sometimes regresses — so the disciplined move is to ship at good-enough and spend scarce human effort only where it actually converts. The source's framing: "AI is getting really good at getting you 80% of the way there, but no matter how many times you prompt it, it won't get you that final 20%… there's a chance you might actually see slight regression trying to make it perfect." The load-bearing distinction is that this is not a licence for slop — "good is not slop" — it is the recognition that returns on prompting diminish sharply near the top, so re-prompting to chase perfection is wasted effort at best and quality-destroying at worst.
Two rules follow. The 80/20 allocation: 80% of a task should take ~20% of the time — lean on AI to reach the asymptote quickly — while the 20% that matters takes ~80% of the time, and that is where you spend human energy to carry good to great. And the judgment of what needs to be great at all: the best builders "know what needs to be good versus perfect" and deliberately build to the minimum viable standard, because most things do not warrant the last 20% (the source cites Peter Levels: only 4 of 70+ launched projects made money — so shipping good and iterating beats polishing things that will fail anyway).
This is the deliberate reconciliation of a real tension with the "quality over quantity" thesis (see Execution Commoditization): quantity-of-slop erodes value, and perfectionism wastes it — both are failures of allocation. The resolution is that quality means choosing which 20% to make excellent, not making everything perfect. The asymptote claim itself is an observation about the current generation's behaviour on the operator's tasks, not a proven law — where the ceiling sits, and whether it is 80% or moves with the model, is not measured here.
Claims
- Returns on prompting flatten near the top: AI reaches "good" quickly but the last stretch to "great" resists further prompting and can regress. observation — asserted about the current generation on the operator's kind of work, not proven; the exact ceiling ("80%") is illustrative, and it plausibly moves with model capability. Recorded as the source's claim.
- Ship at good-enough and reserve human effort for the last 20% that prompting can't reach — chasing perfection by re-prompting is wasted or counterproductive. best practice — context: work where "good" clears the bar and the marginal polish is low-value; explicitly not for the cases where the last 20% is the whole point (safety-critical, signature craft, the thing your brand is judged on). The skill is telling those apart.
- Good is not slop — the discipline is choosing what needs to be great, then building the minimum viable standard for the rest. principle — durable: it resolves the apparent clash with quality-over-quantity by relocating "quality" from everything to the deliberately-chosen 20%. Perfectionism and slop are the two ways to misallocate the same effort.
- Apply the 80/20 split: 80% of the task should take ~20% of the time (AI to the asymptote), the 20% that matters takes ~80% of the time (human, good→great). best practice — context: time allocation on AI-assisted deliverables; a heuristic, not a measured ratio.
- The video states Peter Levels launched 70+ projects and only 4 made money and grew (~95% failure), as evidence for shipping good and perfecting later. observation — attributed, groundable, check-worthy; recorded as claimed.
Related
- Execution Commoditization — the tension partner: that page says AI makes producing free so quality is the premium; this page says perfectionism wastes that premium. Both are resolved by allocating quality to the chosen 20%, not to volume and not to everything.
- Attention Budget — the "where to spend it" companion: the last-20% polish is exactly the high-judgment work the attention budget should be reserved for, not the 80% AI already delivers.
- Evidence-Gated Completion — the guard against reading "good enough" as "done": the asymptote says stop polishing, evidence-gating says prove the good-enough version actually works first.
- Agentic Simplicity — "good is good enough" is the output-quality sibling of "simplest thing that works"; both refuse effort that doesn't demonstrably improve the outcome.
- Imagination Constraint — if most projects fail regardless, the leverage is in which to build (task-selection), which is why shipping good and moving on beats perfecting one bet.
- Distillate: You're the Problem, Not Claude — Six Fixes to 10x Output