Hello! 👋

I recently got back from my first ever trip to the Bay Area! Took the fam out early so we could visit Monterey, Yosemite, and San Francisco proper. After sending them home, I attended Temporal Replay 2026 at Moscone South. I wrote up my experience to share with my colleagues, and now you!

How Temporal Replay 2026 changed my thinking about Temporal at Convergint ~ chadxz.dev

Replay clarified why Temporal is worth pushing further at Convergint and reminded me to support adoption without building ahead of demand.

What’s been going on

Adapting to the new world of software engineering with AI 🙃

After joining Convergint as a Principal Platform Software Engineer, I felt duty-bound to keep close tabs on the AI scene and help my colleague engineers harness it fully. We’ve been using a mixture of Codex, Claude, and Cursor, but most of our engineers seem to prefer Claude, despite the frequent outages. I’ve been moving between different harnesses and models to get a feel for them, but right now I’m using Cursor’s new Agent UI w/ GPT 5.5-xhigh as my daily driver. I’ve been feeling some pain around Git worktree initialization that I should be able to solve with Cursor’s built-in functionality, but if I can’t, I may try out Conductor - a fan favorite around here.

Challenges with writing

Ever since I started using AI heavily in my coding workflow, I’ve felt a pull to also have it do my writing as well. While this does tend to work pretty well for things like writing pull request descriptions or comments, it falls on its face for more prose-style writing. Notion documents, RFCs, and blog posts feel soulless, resulting in text nobody, including myself, wants to read. I’ve had to back WAY off from using AI in these cases, only using it to help me draft outlines and collect reference material. I’m also beginning to push back on others’ writing as well, mostly asking for more succinct documents.

AI on our platform?

It often comes up how we are bringing AI into our engineering platform we are building at Convergint. So far my stance has been to support it heavily in the engineering SDLC for individuals, but I’ve only begun to sprinkle it in places where I feel like it could lift some burden. For example, I have built a “Dependabot Auto Review” shared GitHub Action that runs a Dependabot Pull Request through a model to help determine if it’s safe to merge, or needs additional work. I’m also working closely with Datadog to experiment with new features they are working on, particularly their Bits AI functionality that is meant to automatically draft Pull Requests when an error is detected that it thinks it can fix. Our engineering organization is using AI some for doing things like translations, but we’re not building actual AI-powered features into anything just yet, so there hasn’t been much demand for platform AI support.

What I’m reading

I recently finished Frictionless by Abi Noda & Nicole Forsgren and can definitely recommend it. It gives a practical playbook for building and operating a product-minded platform engineering team, which is great supplemental material for bringing new platform engineers up to speed on the practice. I just started reading Hands-on Large Language Models by Jay Alammar and Maarten Grootendorst, two visual educators in the ML/AI space. I’m only a little ways in but feeling somewhat out of my depth already, so I’m supplementing it by asking ChatGPT for help understanding concepts as I go. On deck for reading is AI Engineering by Chip Huyen and the second edition of Designing Data-Intensive Applications.

Let me know what you think of the article!

Whether you enjoy the article or not, do let me know what you think! You can reply to this email or join the conversation on LinkedIn or Bluesky. Feel free to DM me, too.

May your prompts be one-shot ✌️,

~Chad