Mid-year I wrote that plenty of threads were still running. Q3 is where most of them landed.
The AWS landing zone is live — and I designed and built it myself. Architecture through to hands-on implementation, end to end. Account structure, guardrails, identity model, centralised egress. No delivery partner, no vendor doing the heavy lifting. There's a particular clarity that comes from being the same person who draws the diagram and then has to make it real. Every shortcut you're tempted to take, you meet again three weeks later.
GitHub Enterprise is standing, with Copilot rolled out organisation-wide. Not just procured — configured, governed, and put into people's hands. AI-assisted development stops being a talking point the moment it's in the default toolchain.
Applications are deployed and running in the landing zone. I architected the solutions, then built the harness the developers actually work in — AI-augmented, wired to CI/CD pipelines out of GitHub Enterprise. The deployments weren't the interesting part. The interesting part was designing the path so the next team doesn't need me walking beside them to use it.
WAF migration to Cloudflare, completed. Cloudflare now sits in front, wired through to the landing zone's centralised egress. One coherent edge story instead of three partial ones.
Four things, one quarter, one architect. I'll be honest — it was relentless.
What's next.
The build phase is largely done. The next phase is harder, because it involves other people.
Migration. Driving and supporting the app teams as their systems move into AWS. The platform exists; now it has to be adopted. That's a different discipline — less architecture diagram, more listening to why a team is hesitant and solving the thing they're actually worried about.
AI-driven solutions and architecture. Not AI bolted onto existing systems, but architecture that assumes AI in the design from the start. This is where I think the genuinely interesting work of the next few years sits.
Enterprise architecture practice. Laying down the standards, patterns, and guardrails so good decisions become the default rather than the exception. Architecture that only lives in one person's head isn't architecture — it's a dependency.
TOGAF, still on the horizon. Winter target holds. Time remains the scarce resource.
Looking back at the mid-year post, what strikes me is how much of 2026 has been about foundations — the unglamorous, load-bearing kind. Landing zones and CI/CD pipelines don't make for exciting demos.
But runways aren't built for their own sake. They're built so other things can take off.