Thought Garden stays a calm companion
The rebrand leans into planting, tending, and harvesting progress without turning the product into a gamified farming app.
Living notes
Open questions and assumptions stay visible in the product.
This page is intentionally part of the app so product ambiguity and implementation tradeoffs stay explicit while the concept gets sharper.
View saved source transcriptAssumptions made
The first release is a web app first
The transcript hints at mobile app behavior, but this implementation starts as a responsive web app deployed on Vercel.
Calendar behavior stays read-only in v1
Thought Garden can later suggest placements or all-day task overlays, but this build does not modify Google Calendar.
Document support starts with text-like files
Plain text, markdown, JSON, and CSV are treated as supported first-wave dump formats. Rich document extraction is deferred.
Audio parsing is transcription-first
Uploaded audio and direct in-app dictation are transcribed through OpenRouter before the dump parser structures tasks from the resulting text.
Files are stored in S3 and expected to be served via CloudFront
The app issues S3 presigned upload URLs and constructs file URLs from a CloudFront base URL when configured.
Open product questions
Should Thought Garden ever write to Google Calendar?
Jeanna wants one place to understand the day, while Marra wants the real calendar untouched. The exact integration boundary still needs a product call.
What should the priority interaction be in v1?
The transcript references swiping, pairwise 'this or that', and a March Madness style bracket. The first implementation only scaffolds the data model.
How aggressive should reminders be?
There is interest in daily briefs and sharper accountability prompts, but tone, channel, and timing are still undefined.
How should blocked tasks be handled?
Jeanna described a 'funk buster' flow for low-energy tasks. That behavior is not designed yet beyond the idea capture here.
Which document formats are mandatory for launch?
If DOCX, emails, voice memos, or image OCR are important, the parsing pipeline and storage policies will need to expand.