andershaig

I'm going a bit broader with my system but it's similar. Instead of trying to predict future behavior, I'm focusing on capturing new, meaningful information, preparing it for human review (no fully automatic writes) and pairing with an ongoing review process, deduplication and treating it like a resource I manually curate (with help to make that easier of course).

I think context needs to be treated primarily globally with the addition of some project specific data. For example, my coding preferences don't change between projects, nor does my name. Other data like the company I work for can change, but infrequently and some values change frequently and are expected to.

I also use Claude Code a lot more for thinking or writing these days so there's less separation between topics than there are in separate repos.

I have one of these - it's awesome, I love it and I think it's an incredible success for a first hardware product. My takeaway from reading this is that caring about building a great product made a huge difference in how your first 500 units landed. Now the next batch gets to come with all those learnings.

I'm sure there will be more challenges as well but as long as you keep focusing on the experience you're delivering, I'm sure you'll continue to get past them

It's clearly labeled as the output of Sonnet 3.7, not truth. We all need to apply our own critical analysis of anything we read, whether it's claimed to be from an LLM or from Wikipedia. The possibility of inaccuracy isn't a reason to withhold comment.

Absolutely. I was surprised both in the scale of spam attempts in certain scenarios and how quickly it died with different mitigation measures. It's a challenging thing to blog about because some of the heuristics can be fixed. To be super vague, when you have a certain amount of data about a user if metadata A should be correlated to metadata B and it isn't, that bumps the score. It's not enough on it's own if there are legitimate reasons it doesn't have to be correlated.

I'm always happy to chat through some of the details individually.

I'm purchased a business (kwesforms.com) to follow my dream of being my own boss and right now I'm working on learning. I've built so many different pieces of software but never been in a position where everything is new, I'm playing every role (learning dev ops, marketing, sales etc too) and I have full control to make it the best it can be.

The biggest piece thing I'm working on is bringing some automated form generation tech I've been working on as a side project (ottoiq.com) to it.

I'm also working on a little AI tool to turn todos into actionable plans by gathering additional context and breaking it down. If anyone is interested in talking about their mental models for problem solving, my email is me@<my name>.com.

I can't remember the details anymore but the craziest bug I ever found was one that would crash the page whenever Chrome's Developer Tools were opened. I ended up having to use an embedded JS-based dev tool to figure out what was going on (I was a pretty junior developer and didn't have any better ideas at the time).

I've started tracking an index of projects and their status in Notion. Then I create an extra page based on a template to put things on hold but make them resumable in the future.

In my index, I track: name, status (active, paused, inactive), description, goal and a link to the archive doc if it exist.

My archive doc looks like this (I generally delete any sections that aren't relevant to keep these easy to create):

  # <TITLE>
  ### *Handoff to Future Me: <project name>*
  ### *Snapshot Date*: <date>
  ---
  ### *Project Summary*
  - *Objective*: Briefly state what you're trying to achieve.
  - *Motivation*: What drove you to start this project?
  - *Current State*: Is it in the ideation phase, research phase, or have you already built something?
  ---
  ### *Essential Context*
  - *Related Projects or Dependencies*: Are there any other projects or tasks that are connected to this one?
  - *Technical Specs*: For example, in your lamp project, what type of metal, what voltage for the lamp, etc.
  - *Non-Standard Tools & Environment*: Any unique or specific tools you're using. For example, a specific code IDE or a special type of screwdriver.
  ---
  ### *Progress and Milestones*
  - *Last Completed Milestone*: What was the last significant thing you accomplished?
  - *Next Steps*: Like you said, for your lamp: research, clean metal, buff, etc.
  - *Stumbling Blocks*: Any challenges you foresee or have encountered?
  - *Any Experiments/Tests Conducted*: Brief on what you've tried and the outcomes.
  ---
  ### *Resources*
  - *Important Files & Locations*: Where are the project assets or codebase stored?
  - *Reference Material*: Links to guides, tutorials, or papers that are crucial.
  - *Key Contacts*: Who can you consult about this? Even if it's an online community.
  ---
  ### *Handoff Summary*
  - *Why Stopping*: Why are you pausing this project?
  ---
  ### *Notes to Future Me*
  - Personal notes, reminders, or advice to your future self about the project.