Brian McKeown

Code Story · S13 Bonus: Eliminating Paperwork & Manual Data Entry in…

• What Mistake Did Harmony AI Make?

  Brian McKeown says one mistake at Harmony AI was prioritizing speed over quality. At an earlier deployment, the team built the software fast, then realized the behavior was not the way to deliver what the customer actually wanted.

  Noah Labhart asked Brian McKeown for a mistake and how he and the team responded. Brian McKeown ties the story to the speed-versus-quality trade-off. The work was for an early customer, so he says it is hard to call the choice a pure mistake. Downstream, the software already behaved one way. As Harmony AI filled in missing pieces and gathered more context on the solution the customer wanted, the original approach no longer fit the larger picture.

  The response is

  ▶ at 28:02 Share Transcript

• How Does Harmony AI Scale?

  Harmony AI scales by targeting mid-market, high-volume, family-owned manufacturers and sticking to that market at the start. Brian McKeown says a solution delivered at one factory can be ported to another and connected into the data model, so implementations and deployments get faster one after another. He calls the result a snowball effect.

  Brian McKeown says there is a tendency at startups to accept anyone as a customer to grow revenue and look big enough to go to San Francisco and raise a big, crazy next round. He says a short-term influx of capital might not be worth taking someone outside the ICP on as a customer. Even 5, 10, or 15 more customers, when each one does not fit the ICP and

  ▶ at 24:19 Share Transcript

• How Does Harmony AI Decide What to Build Next?

  Brian McKeown says Harmony AI decides what to build next from the people living in the weeds of customer factories. The team holds product roadmap meetings, probably weekly, and asks what problems show up in a factory and how a problem could translate to someone else's factory. The meetings produce a bird's eye view of how manufacturing behaves.

  Two weeks before the conversation, Brian McKeown was at the International Woodworking Fair in Atlanta. A woodcutter with a sawmill in West Virginia said customers such as Mossberg and Moon Pie do not seem to behave like a woodcutter. Brian McKeown asked whether the mill has machines. The mill debarks the log with a machine, runs chop saws in the pla

  ▶ at 16:50 Share Transcript

• What Does Wellborn Cabinets Love?

  Brian McKeown says Wellborn Cabinets, a customer in Alabama, loves guns and treats the Second Amendment as a favorite amendment. The factory has a skeet shooting range and takes customers there to shoot clay. His team gives hyper-personalized custom gifts so customers feel valued, because a gift shows they listened and learned what the person likes, and that helps build a connection.

  During the pilot, the forward deployed engineers worked on site and built a close connection, which Brian McKeown says is part of the job. Once they learned the Wellborn Cabinets team shoots clay with customers, they called Mossberg, also a customer, and asked for a favor. Mossberg agreed to make two custom Mos

  ▶ at 10:35 Share Transcript

• How Does Harmony AI Close Every Factory Pilot?

  Brian McKeown says Harmony closes a plant by solving the hardest problem in four to six weeks. Every single pilot so far has closed.

  The trade-off is speed versus quality. He wants to enter a plant and move as fast as possible. The pilot doubles as a sales tactic. The team says they will spend four to six weeks on the hardest problem, then do it even when the plant expects failure. Customers laugh and call the confidence excessive. McKeown says Harmony is batting a thousand, because every pilot has closed. The same four to six weeks still have to produce something extremely useful and avoid a shoddy foundation once the contract closes. Nobody wants to repeat the pilot work after the deal is

  ▶ at 8:45 Share Transcript

• How Does Harmony AI Connect a Factory?

  Harmony AI is an AI operating system for a factory. Brian McKeown says Harmony sends a person into the plant for anywhere between three to nine months. The job is to digitize paperwork on the floor, connect machine data, and link the internal tools the plant already runs.

  The on-site work pulls in the ERPs, the CMMS, and the MES, the stack of software tools a factory already has, and connects them into one data layer. Harmony calls the layer an ontological data layer. It is the foundation for AI production scheduling, quality intelligence, and inventory intelligence. The same foundation can automate back-office workflows with AI agents, the tedious work that currently occupies a plant's bes

  ▶ at 3:51 Share Transcript