I researched code agent frameworks including Factory and Cognition, tested multi-modal generators like Flux and Suno, discovered DocxTemplater for document automation, and experimented with the browser's File System API using showDirectoryPicker.
I learned about LLMs simulating Doom in real time via GameNGen, Val.town's dynamic image generation prompts, and Cursor's speculative edits. I also found that ChatGPT currently outperforms Flux.1 for creating consistent cartoons and comic strips.
New image models are getting much better at comic-style generation, especially with text rendering and prompt expansion, though model quality still matters more than prompt detail alone.
I explored techniques for managing LLM coding agents, Anthropic's multi-agent architecture, and persona vectors. I also found handy tools like gitingest for repo ingestion, the O*NET database for job analysis, and modern browser APIs for file access.
I explored using LLMs for educational games and market research, compared log storage costs across cloud providers, and tested document conversion tools. I also found that NumPy often outperforms HNSW indexing for similarity searches on datasets under 1M vectors.
I explored using Pyodide for browser DOM access and PyMuPDF4LLM for converting PDFs to Markdown. I also compared AVIF to GIF compression, experimented with Opus audio encoding, and researched Anthropic’s contextual retrieval methods for improved RAG performance.