Taylor Bosier

Capacity: committing before you know demand

Contract volume has to be committed for the quarter before anyone knows what the quarter holds. Commit too little and you buy the rest on the spot market at whatever it costs that week. Commit too much and you pay a shortfall fee on trucks you never used. The fleet is finite, so every load you promise one lane is a load you cannot promise another.

A two-stage program with recourse over 400 demand scenarios. The capacity constraint binds only the first stage, so the second stage separates by lane and the Lagrangian decomposes. One price on capacity goes out to every lane, each answers with its own newsvendor quantile, and the plan is where those quantiles meet the fleet. The last step prices it against planning on mean demand, and against knowing demand in advance.

Pricing: quoting a lane that barely runs

A lane with five loads should not be quoted with the same confidence as a lane with a hundred, but it still needs a price today. The question is how much of the network's rate a thin lane should borrow, and how much of its own history it has earned the right to trust.

A three-level Bayesian hierarchy settles it. The shrinkage weight below is how much of the quote came from the lane itself. Click a lane, or use the arrow keys, to price it.

Dispatch: pickup and delivery routing

Eleven orders, four trucks, and nothing can be delivered before it has been picked up. Finding a set of routes that works is easy. Finding a cheap one fast enough that a dispatcher will actually wait for it is the job.

220 candidate routes are built by randomized cheapest insertion under precedence, capacity and range. A greedy pass then covers the orders with the cheapest disjoint set it can build. When the fleet cannot reach all of them, the readout says how many it served. It is a heuristic: no LP relaxation, no dual bound, no gap. In production that is what a solver is for. Switch to Data for the instance an operator would receive, or Assignment to see drivers matched to chains of orders.

Agents: one deduction case, worked twice

A customer short-pays an invoice and someone has to find out why. Pull the data, check it against the contract, ask whoever knows, write it up. Nearly all of that is looking things up. The part that needs a person is a question or two for a manager.

On top, an analyst agent works the case. Its first tool call goes to a SQL data agent, which reads a knowledge base on what the data means before it writes the query. The analyst agent then runs its checks as parallel tool calls. When the price will not reconcile, it messages the associate manager on Microsoft Teams through Azure Bot Service, waits for the answer, and writes what it learned to a knowledge base that a vector search index reads, so a later case like this one can look it up. Underneath, a person works the same case by hand: a Power BI report, an export, a workbook, a message and a wait. Then the view pulls back to a whole queue.

Drawn, not measured. This is a drawn walkthrough. The customer, the case, the prices and every elapsed time are invented, and the timings are my guess at the shape of the work. In both panes the manager answers just as fast, and in the queue every case waits on the manager. The queue shows the idea the agents are built on, one agent carrying a queue that takes a team by hand. The agents this is drawn from are designed and through security review but not deployed, so there are no results yet. Read the counts as a picture of that idea, not a measurement.

Contract intelligence: documents to decisions

The rebate terms are three paragraphs buried in a sixty page contract, and a contract estate can run to many thousands of them. Reading all of them into a model costs a fortune and still misses things, so the job is to find the two paragraphs that matter and price the rebate off those.

Below is one contract going all the way through: it arrives, gets read, lands in the lake, is chunked into a vector index, and an agent pulls back only the clauses it needs. The rebate owed falls out at the end.

None of this is real data. The contract, the customer and every figure on it are invented, and the percentages are arithmetic over those invented figures. It is a drawn walkthrough of the shape of a retrieval pipeline I built in industry, not a window onto one.