calebwoods

Solving business problems with technology. AI and custom software development. CEO @rolemodelsoft. Christian, husband, software craftsman.

[calebwoods @calebwoods]

The businesses getting real value from AI aren't getting it by accident. They're working from a system.

We wrote ours down. Four free guides, personal practice to team capability:

Collaborating with AI
AI Skills
AI Tooling
AI Workspace

https://t.co/qcjKUy8XRm

[calebwoods @calebwoods]

RoleModel's full guide: https://t.co/6LjlRGsZTL

[calebwoods @calebwoods]

I run all of my AI work out of one local project. A workspace. Claude and Codex both operate in it. Same files, different tool, nothing rebuilt.

Providers will keep changing. That should cost you a subscription, not the thing you have been building all year.

[calebwoods @calebwoods]

Past a point, more instruction makes AI output worse.

Refine is the step people get wrong, because it does not just mean add more. Something comes out badly and the instinct is to write another rule.

RoleModel's full AI Skills guide: https://t.co/05Ouu1QLpE

[calebwoods @calebwoods]

"What problem are you trying to solve?" is a better question in a world of AI prototypes. It is also the easiest one to skip.

The risk isn't that the prototype is wrong. It's that it works well enough to stop the conversation.

[calebwoods @calebwoods]

A fresh chat asks generic questions.

A model that holds your prior work asks better ones, then weighs your answers against decisions you already made.

The prompt is one line anyone can copy. The context isn't.

[calebwoods @calebwoods]

If you need the same answer every time, stop prompting for it.

Have the model write the script. Then run the script.

The LLM is the prototype. What proves itself gets hardened.

[calebwoods @calebwoods]

Published a new blog post from my learning on toggling offline mode during tests with Selenium/Chromedriver https://t.co/rOMbMQD3Dp

[calebwoods @calebwoods]

My post on the @rolemodelsoft blog about an alternative to estimates is live https://t.co/vo0a1QBdCr

[calebwoods @calebwoods]

Blog post with part of my story of coming to @rolemodelsoft and the impact of Kevin Bridges as a mentor in my life and career https://t.co/jBJHdwUllL

[calebwoods @calebwoods]

After experimenting with fasting for the last 6 months, decide to write up my experience. An easy implement, high reward New Year's resolution. https://t.co/8TicsNMQFI

[calebwoods @calebwoods]

Wrote a blog post on learning Craftsmanship through Production Support experience https://t.co/Ix75GBZHsz

[RoleModel Software @rolemodelsoft]

Why production support experience is key to your software project's success: https://t.co/CXuGD7tWur by @calebwoods

[calebwoods @calebwoods]

@Dodds_M Biggest question for me is how modeling a domain would work in Elixir's functional paradigm.

[calebwoods @calebwoods]

@Dodds_M we've started doing some experiments @rolemodelsoft. Don't have any solid opinions yet. Seems promising.

[calebwoods @calebwoods]

@_viniciuspalma_ I had to change the jasmine-rails config file to support it. Might also still be an open issue on jasmine-rails github

[Raspberry Pi @Raspberry_Pi]

https://t.co/GwijoS4ros

[calebwoods @calebwoods]

@moeamaya glad you found it helpful.

[calebwoods @calebwoods]

Summarized my recent thinking about naming projects in a blog post: https://t.co/8vckBdxPuS

[calebwoods @calebwoods]

Been running f.lux on my iPhone for a couple months and loving it, although an iOS solution should be more polished. https://t.co/DFHvMuOcLE

[Noel Rappin @noelrap]

Looks like we now know why Apple was giving f.lux a hard time on iOS: https://t.co/pz7m9bw62B

[calebwoods @calebwoods]

@jackboxgames keep getting games reset mid round playing FibbageXL on the Apple TV tonight. Known issue?