Why off-the-shelf workflows run in weeks and novel ones take far longer — the "march of nines."
It depends almost entirely on novelty. An off-the-shelf workflow on proven hardware can be running in days to weeks. A genuinely new workflow takes much longer — not because the first version is hard, but because reliability is: getting to 90% is quick, and each additional nine (99%, then 99.9%) takes roughly as much effort again. Budget for the nines, not the demo.
The single biggest predictor of deployment time is whether you're running a known workflow or inventing one:
A demo that works 90% of the time looks finished. It isn't. Each extra nine of reliability — 99%, then 99.9% — tends to cost about as much effort as everything before it, because you're now handling the rare edge cases that only appear at scale.
New workflows drift because new requirements keep appearing — an edge-case sample, a reagent that behaves differently, a step that needs a tolerance no one specified. The first 90% is the part everyone imagined; the last stretch is the part nobody did. That's normal, and planning for it is the difference between a deployment that lands and one that stalls.
