Split a job across several model calls, derive from arithmetic how the shape you chose multiplies errors and adds latency, and be able to show whether the split beats one careful call on the same work

Many Models, One Job

Pipelines, fan out, loops and routers priced honestly, with the error arithmetic that decides whether splitting a job helps or hurts.

8 lessons, written and corrected before you arrived. Reading them here needs no account. The first reads the whole way through; the others open and then stop, because a page nobody owns cannot tell who is reading it. Starting the course gives you your own copy, where every idea has problems standing under it and you can ask about any sentence.

Start reading

01What Splitting a Job Across Model Calls Buys, and What It ChargesFour model calls in place of one doubles the price and more than doubles the wait, so each boundary has to earn that, and only three kinds of boundary do.
02A Step Good 95 Times in 100 Is Not Good Enough to Be the Fourth of Fouropening onlyWhere the product rule for a chain of model calls comes from, what four steps at 0.95 actually deliver, and the two things that can be done about it.
03Chain, Fork, Loop and Crowd, Priced on One Jobopening onlyFour ways to arrange several model calls behind one request, each drawn and each priced on the same order-answering service, with the two numbers that decide between them.
04Who Decides the Next Step, and What Deciding Costsopening onlyA router is a step with an accuracy of its own, and it multiplies everything behind it. The arithmetic for choosing in code against choosing with a model call, on one order-answering service.
05What One Step Has to Tell the Next, and What It Must Notopening onlyThe handover between two model calls is an interface, and carrying the whole history forward instead of a fixed set of named fields nearly doubles what a four step run costs.
06What a Gate Catches, What It Destroys and What It Costsopening onlyA gate between two steps cuts the bad share of handovers from 15 per cent to 5.3, and the same four numbers show what it destroys, why it loses money on compute, and where to set it.
07The Loop That Will Not Settle, and the Loop That Settles on Something Wrongopening onlyA retry loop that fixes a draft six times in ten takes 1.56 passes on average, and the cap you put on it moves the worst case rather than that average.
08Three Points of Accuracy for Twice the Costopening onlyThe four step version is right 81 times in 100 against 78 for one call, at twice the cost and 2.3 times the delay, and whether that is worth buying is arithmetic.