Joseph Hale

@thehale.dev

397 Followers 121 Following

Software Engineer, Musician, Speedcuber | @37signals | ¡Hablo Español!

Solving a Rubik's Cube in 11.034 seconds at Pier 39 in San Francisco
Two Python code snippets demonstrating the value of the new `frozendict` immutable type coming in version 3.15 Typically, you should avoid using a `dict` as a default argument in a function because it's mutable. When 3.15 is released, you can safely use a `frozendict` instead.
[LinkedIn Post by Ben Bradford] Every time I switch AI tools, I start from zero. I'm starting to think that I can't be the only one feeling this frustration. Here's a small example...I recently used an AI tool to build a presentation and the AI kept structuring slides in ways that didn't match how I actually present. It was text heavy, wrong flow, missing my voice. It had no idea how I work because it had no context about my previous presentations. I stopped and built the slides manually. My problem is that my context is locked up somewhere else. I've spent hours with Claude, ChatGPT and Gemini and none of that travels with me when I jump into a different tool. And this is happening everywhere now. Most of the SaaS apps I use daily are adding AI powered features by the same foundational models. But when I use those features, they start from zero. They don't know anything about how I work or what I need. What I actually want is control over my own context, locally. Build it once, keep it with me, and selectively share pieces with whatever tool I'm using. When a new model comes out that's better for my specific task, I want to test it without rebuilding my entire workflow from scratch. When I open an app with AI built in, I want to drip in just enough context to make it useful without explaining myself all over again. Right now it works backwards. The tools own my memory. I'm stuck rebuilding context in every new app, every new conversation, every new project. And when the AI doesn't have that context, it's not useful to me at all. I've been researching tools like Mem0 and SuperMemory but they feel more like tools an engineer would use to build an app. I know folks like Dharmesh Shah of HubSpot have written about this and he seems optimistic about where the solutions are headed. Has anyone found a better way to manage context across multiple AI tools?
10 lines of sample code showing how to use @thehale/react-native-keep-awake to prevent a phone screen from going to sleep.
A screenshot of Excel's formula evaluator, showing a variable being resolved in-place as part of a step in the evaluation.