I’ve added three user-defined modes to the Echo app—dev, small business, and personal user—each with its own data sets and summarization prompts. The app switches between these modes automatically and tags events with the right OpenAI tag lists, which users can also modify if needed. Fixing bugs around these modes was a big focus; there were issues where custom profiles would overwrite each other’s data in the config file, and problems with resetting configurations or editing user modes after restarting or closing the app. Those are mostly sorted now.

I also improved the GUI for the Notion integration and refactored all user-editable settings and dialogs to make navigation easier. Added a button to measure tokenization rate using Ollama, but I still need to figure out a good way to handle cases where users don’t have Ollama installed—some kind of security barrier or fallback is necessary to avoid errors. The config document is still pretty messy because of the Python interface, so I want to clean that up eventually.

Users can now edit their license info anytime during app use, and I added a license button plus a legal notice on the about/help page. On the backend, I simplified validation checks, refactored how configurations load, and adjusted model selection so it rewrites models based on the active user mode.

Looking ahead, I’m debating whether to add custom profiles and models, but I want to discuss that with the team first. I’m also planning to refactor the menu and logic for first-time setups and create a separate key script to handle health checks and unlocking different modes. Integrating the Obsidian API is on my radar, though I’m still figuring out the technical details. Another idea is adding a tokens-per-second detection feature to help with the energy and cost savings calculator, but that’s still in the early thinking phase.

I’m considering a student mode for transcribing classes or lectures, which could be useful but I want feedback before committing. Packaging version 1 of the app for the first product launch and setting up a website to share it are next steps. I also want to get GPT to write a simple script for a transcript speed function to help with that.

There are still open questions around verifying that modifying the active tag list properly accesses all tag files, how to make the interface less cluttered and more intuitive, and how to implement the TPS detection feature. Overall, progress is steady but there’s a fair bit left to clean up and clarify before the first launch.

Two-Line Takeaway

Key step forward

The most important forward progress step is the addition of three user-defined modes with automatic switching and tagging, along with resolving major bugs related to these modes.

Significant barrier or risk

The most important barrier or risk is handling cases where users don’t have Ollama installed to avoid errors, as well as cleaning up the messy config document and clarifying open questions before the first launch.

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