Skip to main content
Capture by voice or photo on your phone. Bower transcribes, extracts, and syncs to your desktop — so nothing falls between the bench and the write-up.

​
Voice transcription

Bower records your spoken observations and transcribes them in real time using scientific speech recognition trained on domain vocabulary. The transcript appears as an editable note, with the original recording attached.

​
To record a voice note

1

Open Bower on your phone

Go to app.bowerlabs.ai in your mobile browser. For faster access, you can add Bower to your home screen.
2

Tap Voice Recording

Select the Voice Recording tile on the Quick Capture page.
3

Speak your observation

Bower listens and transcribes in real time.
4

Tap stop when you're done

Recordings can be up to 30 minutes long.
5

Your note is created

Bower generates a title automatically using AI, or you can set one manually. The transcript syncs to desktop within seconds. The original audio recording is always attached to the note — so the source of truth is never lost.
A few things worth knowing:
  • Bower recognises scientific terminology: units, chemical names, numeric values, and Latin binomials.
  • Recordings have a 30-minute limit. For longer sessions, use live agent mode.
  • If a value looks outside an expected range, Bird will flag it after capture — not during.
  • Accuracy improves as Bower learns your terminology and accent over time.
  • You can rate the quality of transcriptions and text extractions. Both positive and negative feedback helps us improve accuracy for scientific content.

​
Photo text extraction

Point your camera at a handwritten notebook, printed document, instrument display, or sample label. Bower reads the image and extracts the text as an editable note. (This is sometimes called OCR — optical character recognition.)

​
To capture a photo

1

Tap Take Photo on the Quick Capture page

You can take a new photo with your camera, or tap From gallery to select an existing image from your phone.
2

Frame the document, display, or label and capture

3

Bower reads the image and extracts the content

4

The original image is attached to the note with metadata

​
What Bower can read

  • Printed text and tables (approximately 98% accuracy)
  • Handwritten lab notebooks
  • Instrument displays and gauges
  • Mixed handwriting and print
  • Sample storage labels

​
After extraction

  • Extracted tables are inline-editable — correct values directly without retyping.
  • Low-confidence regions are highlighted. Bird will surface anything it wasn’t sure about.
  • Everything becomes searchable immediately.

​
What happens after capture

Every voice note and photo syncs to your desktop workspace via WebSocket — no manual upload needed. Once there, it’s fully searchable, editable, and available to Bird for summarisation and retrieval.