Code Commute

Blowing your budget and getting rate limited both mean you're using more. I'm fine paying when it genuinely helps me do more. But pick a lane: take my money, or rate limit me and make it cheaper. Either way, watch your tool usage. Watch here: youtu.be/NUOAOiT3mAA

Could I write the code myself? Yes. Did I have the capacity? No. Some evenings I just couldn't crank out code on my own. Leaning on AI was how I kept making progress while work had me completely swamped. Watch here: youtu.be/NUOAOiT3mAA

I wasn't even coding during the day and still kept hitting rate limits. I maxed my Copilot budget, got limited, switched to Claude, got limited there too. Eventually the only move left was to raise the budget I knew I'd blow. Watch here: youtu.be/NUOAOiT3mAA

More instructions in your agent file isn't always better. Jam every rule for tests and SQL into context and the agent drags all of it into work that has nothing to do with either. That noise can quietly make the output worse. Watch here: youtu.be/NUOAOiT3mAA

Copilot CLI was the first thing to pull me out of Visual Studio. After a lifetime as a .NET developer living in the IDE, the CLI workflow stuck. What caught me off guard was how fast I hit rate limits. Watch here: youtu.be/NUOAOiT3mAA

Giving an AI agent read access is risky. Write access is scarier. Reading the wrong data is bad. An agent with approval to mutate things can quietly go drop your production database before you realize what it's about to do. Watch here: youtu.be/NUOAOiT3mAA

Feeling behind on a technology? Go build something with it. It doesn't have to ship or even fully work. Deploy something on Kubernetes, try Blazor, spin up a site with Astro. Exposure is the point, not a finished product. Watch here: youtu.be/leZnc8IxhyE

Don't design the thing you're not the expert on. When you lack the depth, pulling experts in early beats winging the design yourself or dumping it entirely. One leaves bad designs, the other leaves you never learning. Watch here: youtu.be/leZnc8IxhyE

This isn't just a manager problem. Tech leads hit it too. If your strategic conversations involve technology and you can't speak to any of it, that's a real signal. It usually means there's a technical gap worth closing. Watch here: youtu.be/leZnc8IxhyE

Want to stay technical in a strategic role? Sit in on design reviews. Even reviews outside the teams you manage are worth attending. Seeing work you aren't directly responsible for keeps your technical understanding fresh. Watch here: youtu.be/leZnc8IxhyE

Don't fully hand off the thing you don't understand. Pull in your expert early and work alongside them, instead of dumping it and disappearing. Otherwise you'll need that person every single time it comes up again. Watch here: youtu.be/leZnc8IxhyE

Trying to stay involved in everything technical doesn't scale. Force yourself into every decision and you become a gatekeeper who blocks things. Leadership means learning how much depth you actually need, then trusting your team. Watch here: youtu.be/leZnc8IxhyE

Staying technically sharp gets harder as your role gets strategic. Going from developer to manager, or even junior to senior, shifts you toward strategy over tactics. The trick is keeping enough depth without losing the bigger picture. Watch here: youtu.be/leZnc8IxhyE

When your technical confidence fades, your role notices. If strategic conversations involve tech and you can't speak to any of it, take the hint. Losing a deep expert on the team is exactly when you step back into the code. Watch here: youtu.be/leZnc8IxhyE

Deferring to an expert once is fine. Doing it every time is a signal. If you never feel confident speaking to anything technical in a conversation, that's your litmus test. It probably means it's time to go brush up. Watch here: youtu.be/leZnc8IxhyE

Part of being a manager is taking random hits so the team can focus. Whether it's a livesite issue or a partner ask, I try to absorb it first so my team can keep building. It doesn't always scale, but it's worth protecting their focus. Watch here: youtu.be/UBuSqdKMXNU

Your manager probably isn't missing the point on purpose. Most of the time they're just missing context. Give them visibility into what you're juggling and they can make a much better priority call. Watch here: youtu.be/UBuSqdKMXNU

The best time to hand off work is right as someone finishes. The cleanest handoffs happen the moment someone wraps their last task and picks up the next. That beats yelling drop everything in the middle of someone's focus. Watch here: youtu.be/UBuSqdKMXNU

Half of context switching is just unclear priorities. When a stakeholder or your manager flags something, ask clarifying questions first. A lot of the time you learn it isn't actually a drop-everything-right-now thing. Watch here: youtu.be/UBuSqdKMXNU

Startups context switch constantly, and early on that's survival. When you're always chasing the next customer, something new always pulls your attention. At a big company with stable services you feel that pressure far less. Watch here: youtu.be/UBuSqdKMXNU