I made some progress improving the auto notes feature by adding a post-processing step that culls special notes when there isn’t enough content to justify them. I also prioritized which auto notes show up based on the user’s mode, but I still need to figure out the best way to tune those selection options so they feel relevant without being overwhelming.

Naming the app ended up taking more time than I expected. The original name, Echo, conflicted with Amazon’s product, so I went through a bunch of alternatives like Tiny Logger, Crystallization, Amber, Amber AI, and Resin before settling on Nucleate. I’m still a bit uneasy about how close “Nucleate” feels to AMD Ryzen when spoken aloud, but it seems like the best fit for now. I also toyed with simplifying the name by replacing “-ate” with “8,” but I haven’t decided if that’s worth it.

On the animation front, I initially tried coding the crystal growth animation in Python, but I wasn’t happy with how it looked, so I switched to Blender. I managed to create a crystal growth animation there, though it still needs some color tuning to get it right. I’m also curious about using Blender’s geometry nodes to grow crystals from the outside in, but I haven’t tried that approach yet.

I wrote a Python script to simulate crystal growth as part of this, but it still needs refinement to feel more natural and visually interesting. It’s clear there’s more work to do before it’s ready to integrate fully.

For the app’s launch strategy, I’m thinking about offering both a free limited version and a premium version, with a launch price in the $25-$30 range for early adopters. The plan is to do a beta launch early next year, aiming for distribution via Gumroad and direct download by February.

One feature I’m considering adding is a to-do list or running tracker that extracts tasks from the note summaries and maintains them in a JSON list with timestamps. The idea is for a back-end language model to autonomously check notes and update this to-do list on a timed cadence. But I’m worried about cache bloat and false positives, so I’m thinking of implementing a confidence level system where the user manually confirms or dismisses tasks before they get added. I haven’t started coding this yet, but I might draft a version of the to-do list script just to test the concept.

Right now, I’ve decided to pause new feature development on Nucleate so I can focus on some other studio projects. That said, I’m still interested in exploring some of these ideas like the to-do list feature and the Blender animations when I get back to it. There are still a lot of open questions around balancing polishing the app versus launching to get feedback, how best to represent Nucleate’s branding through splash pages and animations, and whether Blender’s geometry nodes will be a good fit for the crystal growth effect. For now, I’m stepping back to clear my head and figure out the next steps.

Two-Line Takeaway

Key step forward

The most important forward progress step is improving the auto notes feature with a post-processing step that culls special notes and prioritizes them based on user mode.

Significant barrier or risk

The most important barrier or risk is balancing the app’s polishing and launch timing while managing potential issues like cache bloat and false positives in new features such as the to-do list.

Keynotes

Decisions & Observations

• Decided to add a post-processing step to cull special auto notes when content is insufficient and prioritize notes based on user mode, but needs to tune selection options for relevance without overwhelm.
• Switched crystal growth animation development from Python to Blender; created an initial animation needing color tuning and plans to explore Blender’s geometry nodes for more natural growth effects.

Issues & Next Steps

• Chose the app name “Nucleate” after rejecting alternatives due to conflicts; considered naming the app Resin but acknowledged potential confusion with AMD Ryzen when spoken aloud.
• Planning a launch strategy with a free limited version and a premium version priced $25-$30 for early adopters; targeting beta early next year and final release by February on Gumroad and direct download.
• Considering adding a to-do list feature that extracts tasks from note summaries and updates autonomously via a backend language model, but needs to address cache bloat and false positives through a confidence system with user confirmation; has not started coding this yet but may draft a test script.
• Needs to resolve how to balance polishing the app versus launching to get feedback, define branding through splash pages and animations, and determine if Blender’s geometry nodes suit the crystal growth effect.

Uncategorized Notes

• Developed a Python script to simulate crystal growth but requires refinement for naturalness and visual interest before integration.
• Paused new feature development on Nucleate to focus on other studio projects but intends to revisit ideas like the to-do list and Blender animations later.

Generated with Nucleate by Watchlight Studio — distilled from full transcription.