Todd McPartlin

I work at the intersection of AI, APIs, and real engineering systems. Not theory. Not demos. Practical ways teams are actually building faster.

At what point do humans stop feeling the need to review code? Not because AI becomes perfect. But because statistically it becomes better than the average engineer reviewing it. Feels like we are heading toward a world where: - humans review architecture - intent - boundaries - security - workflows while the actual implementation increasingly becomes machine-generated and machine-validated. Not sure we are there yet. But I also would not have believed where we are today just 12 months ago.

This transparency is exactly what this industry needs. Thanks @ClaudeDevs and @bcherny

New CursorBench results just dropped. Two big takeaways. Composer 2.5 is way better than most people think. 63.2% score at $0.55 per task. Nearly matching Opus 4.7 Max and GPT 5.5 Extra High at 20x less cost. This is insane value. Gemini 3.5 Flash is #10 at 49.8%. Below GPT 5.5 Low. Below Opus 4.7 Low. Google's newest model can't even beat budget tier competition. Composer 2.5 is the sleeper. Gemini 3.5 Flash is the disappointment.

Using modern AI tooling has been a pretty eye-opening experience. Over two weekends I was able to build: - a React Native mobile app - powered by @convex - an admin/backend portal - powered by Convex - authentication - powered by @clerk - realtime data flows - deployment workflows - powered by @expo The interesting part is not just the speed. It is how much cognitive overhead disappears when the platform, backend, and development workflow all feel tightly integrated. A year ago I would not have believed this level of iteration speed was realistic. Now it feels like the new normal. The bigger challenge going forward is not building faster. It is deciding what is actually worth operationalizing, scaling, and supporting long term.

This is great to see. If you are building apps y’all need to take a look @clerk I just used them for the first time with a mobile app and it was so easy!!!

I have not written a significant amount of manual code in almost a year. A year ago I struggled to give up control because the models were not good enough yet. Today my role has shifted much more toward: - architecture - workflows - orchestration - context and direction The coding itself increasingly feels like managing 100 junior developers. AI did not remove the need for engineering discipline. It made it more important.

@mntruell @elonmusk Love it. Keep the improvements coming. #multimodel

Composer 2.5 is a significant step up from Composer 2. This is the very start of our work with SpaceXAI. Hope to have more improvements out soon.

This is excellent news. Keep it coming. Cursor is setting the benchmark… multiple models is the way!!! @cursor_ai

@KaiXCreator Cursor. Biggest bang for your $20. Claude burns tokens fast. Cursor on Auto with great rule context files and will last a long time.

I’m ready to invest $20, which one should I choose? - Claude - Codex - Cursor - Antigravity - GitHub Copilot Which one is more worth it right now?

@BoazWith Great question. Yes, I would prompt “look at sibling repo X” then it used it and leveraged each other contexts.

@ToddMcPartlin That cross-project context is the useful part. Did Codex keep the contracts straight between the RN app and backend, or did you still need a manual handoff doc?

Spent the weekend using Codex exclusively across: - a React Native mobile app - a marketing website - a backend portal What impressed me most was the orchestration and ability to move across all three projects while maintaining context. Cursor is still my primary environment because of: - multi-model support - editor experience - separation between editor and agent workflows But these tools are evolving beyond code generation very quickly. The real shift is workflow orchestration.

The cost of experimentation just collapsed. Ideas that used to take weeks can now be prototyped in an afternoon. But speed alone is not the advantage. The real challenge is knowing: - what becomes a real product - what stays an experiment - what is worth scaling and supporting long term Speed matters. Judgment matters more.

Thanks. It has been a great experience. She is off to a great start. App should be launching soon

I spent a few hours recently getting my wife set up on her laptop to start “coding”. She is a creative business owner, not an engineer. Within a few hours she was: - creating a React Native mobile app she has been asking me to build for the past 8 years   - building out a marketing site for the app   - experimenting with ideas in real time  Watching that shift was amazing. When you pair strong domain knowledge and creativity with the right engineering guidance and AI tools, the barrier to building drops fast. The future is already here. Small teams and individual builders can move much faster than before. That said, this comes with responsibility: - security still matters   - data handling matters   - how you structure systems still matters  AI makes it easier to build, but it does not remove the need to build correctly. If you are a small business owner, find someone who can help you get set up the right way. The leverage is real. Watching her take ideas and turn them into something working that quickly was a good reminder of where things are heading.

@morganlinton I completely agree. Cursor is fantastic because at least you can choose to spend more or not. I personally use Auto mode with great cursor rules and skills 95% of the time. I write code everyday and don’t spend much over $50 per month.

Officially canceling our Anthropic plan, it’s Codex + Cursor for my little 16 person eng team. Anthropic is great for companies that can spend $2,000/mo and up per engineer, but not affordable for us. Codex really upped their game recently, and with GPT 5.5, it’s just so good, and so token efficient. Still using Cursor plenty, my team still looks and reviews a lot of code. But with Cursor, we’ve never hit a limit, and Composer 2 is pretty awesome for most stuff. Testing out Droid as well and see some good early results with Droid + GLM 5.1, but still more testing to do before rolling it out to the whole team. My guess is many more engineering leaders will be sending messages like this. Anthropic makes great stuff but phew, it’s so darn token hungry. My team loves Codex and Cursor, onward!

“Prompt engineering” is quickly becoming “context engineering”. Better outputs increasingly come from: - repo awareness - APIs - architecture context - workflows - memory AI is starting to reward disciplined engineering teams very quickly.

APIs have evolved far beyond simple system integrations. At Command Alkon, an API-first approach has become foundational to how we scale Command Cloud, enable faster product development, and reduce friction between teams and systems. As AI adoption accelerates, this becomes even more important. APIs are increasingly the bridge between platforms, products, agents, and workflows. The organizations that invest in strong API strategies now will be in a much better position to move quickly as AI and automation continue to evolve.

Release earlier than you are comfortable. Get it in front of real users. Get real feedback. Then iterate. Internal opinions are useful. User feedback is what actually moves the product.

OpenAI models and Codex are now available on AWS Bedrock. For enterprise teams, the real news is not which models — it is that you can now route OpenAI through IAM, PrivateLink, and CloudTrail. The compliance blocker just got smaller. What are you going to build now? openai.com/index/openai-o…