there are 3 ways to log:

1. Scan
2. Voice input (transcribed, parsed, compared to db)
3. Choose from a database (search) - kind of tedious and I dont like it.

Scan or voice is my favorite because you can get precise logging. Voice is best imo because you can see exactly what it types as you speak and then turns that into portions that you can adjust.

Scans also show portions which you can adjust, this is powered by a fairly low end LLM to reduce costs.