Where AI is worth doing in the Twin Cities
Manufacturing in the Twin Cities is mostly small shops, and many of them sell to medical device makers and other large customers who audit what they buy. In a machine shop of a few dozen people, every job drags a set of documents behind it: the quote, the inspection records, the certificates, the change notices. Each has to agree with the others and with the current drawing revision, and checking that falls to the quality lead. The insurance agencies and offices around the plants carry a similar load in applications, renewals and follow-up that repeat every day.
For that shop we would build a document checker for the quality lead. A certificate or inspection record is read as it arrives and set against the order and the drawing revision on file. What does not agree goes into a short list for the quality lead, who decides whether a part is held, reworked or released. In an insurance agency, the same kind of tool gathers what is missing from an application, and an agent reviews advice and anything that binds coverage. Before building, we compare a tool the business already owns, ordinary automation and AI, and build the smallest thing that works. Paperwork that agrees is filed to the job without anyone handling each sheet. We run our own businesses this way, with court-record lookups and reminder texts handled by software we built, and nobody lost a job.
Consider an order from outside Minnesota: a buyer in Sioux Falls sends a drawing and asks for a price. The request is logged, compared with what the shop makes and placed in front of the estimator, who sets the price. When the order follows, its certificates and inspection records pass through the same checker as the device maker's, and the buyer hears when the parts have left. The last message in the thread is the buyer confirming delivery. Nobody from the shop drove to Sioux Falls, and nobody was hired to handle the order. One from Wisconsin or Iowa, or from a buyer overseas, would take the same route.



