Episode #459
459- Make the Approved Path Fast Enough w/Brandon Muncy
Everyone's using ChatGPT on their own to save 30 minutes a day, and nobody will wait six weeks for a review, so Brandon Muncy, Director of Information Technology at PepperPointe Partnerships, wants to make the approved path fast enough that a shortcut isn't tempting. After building the dental service organization's in-house IT team from a fully outsourced start, he explains why onboarding broke even when everyone did their job and why he expects the next 18 months to be about identity and permissions for AI agents. Brandon Muncy's path into IT leadership wasn't a straight line. After an engineering degree he went into retail management at Verizon and AT&T to pick up the business side, then moved into research computing at the university. From 2018 to about 2021 he was the Linux person on a 10-node cluster, hundreds of VMs, a deep-learning machine and a VDI setup for researchers. PepperPointe, where he now leads IT, is a dental service organization in Lexington, Kentucky with hundreds of practices across a couple of states. When he joined in late 2022, its IT was fully outsourced, with no internal team. He has built a team of about seven people for upwards of 1,500 users. He found out that you don't change the perception by explaining it. Better decks didn't work, because nobody updates a belief based off a slide, so he works one visible win at a time, and onboarding is one of the vehicles. When a new hire starts on Monday and IT finds out Monday, everybody has a bad week, even though everyone involved did their job. Instead of blaming the office, which is busy running a practice, his team went to the office managers and, looking at onboarding from their chair, found a defining step that tells IT sooner. He also describes his real bad day: by 4 o'clock he has been useful maybe 11 times and hasn't moved a single thing forward. On AI, Brandon says that in healthcare it shows up as a governance question before it shows up as a capability question. People using ChatGPT on their own save 30 minutes a day and won't wait six weeks for a review, so he wants an approved path fast enough that a shortcut isn't tempting, even if the first version is half-baked. Only his administrators have access to Copilot, so the team can start working on DLP policies. His 18-month prediction is identity and permissions for non-human actors: if the tool touches patient data, it's a business associate, and he expects an enforcement story. Key takeaways: Make the approved path fast enough that a shortcut isn't tempting, even if the first version is half-baked.; Change how the business sees IT one visible win at a time, because nobody updates a belief based off a slide.; When onboarding breaks, look at the process before you blame the office, then give the office one defining step that tells IT sooner.; Keep Copilot to your administrators so your team can start working on the DLP policies.; Guard your strategic thinking time: a day of being useful 11 times can end with nothing moved forward.; Treat documentation as part of the process, because it loses the prioritization fight every time.
