Voice or text journaling that lands in your Obsidian vault, linked and searchable
How it works
The user talks or types. If it's voice, the agent sends the audio to NEAR AI's private openai/whisper-large-v3 endpoint at https://cloud.near.ai/models/openai/whisper-large-v3 for transcription — no third-party cloud, audio never leaves NEAR AI infrastructure. The agent then asks two setup questions on the first run: 1) where to save (which Obsidian vault or local Markdown folder, e.g. ~/Documents/ObsidianVault/Journal/) and 2) how much structure per entry (minimal: just timestamp + tags, standard: title + date + tags + mood + 1-line summary, or full: standard + key insights + action items + linked entities). The agent remembers the choice and applies it to every future entry. On every save, it auto-tags from the content, detects recurring entities (people, projects, companies, ideas), and inserts Obsidian wiki-links [[like-this]] to any existing note that matches.
Every night it generates a short daily digest entry that links out to the day's journal notes. The user can retrieve entries by exact keyword (filename or in-note text) or by semantic query ("the conversation I had with my co-founder about Q3 pricing").
Example Prompt
“I just had a long talk with my co-founder about our Q3 roadmap — record it as a journal entry, link it to my existing "Q3 planning" note, and pull it up later when I ask "what did we decide about pricing last month?"”