From whiteboard to action plan in about ten seconds.

## What happens in those ten seconds

0–1 second — Capture and correction. You tap the shutter. Apple VisionKit runs on-device, straightens the perspective, and produces a clean cropped image of the board. This step happens entirely on your iPhone's Neural Engine — no network involved.

1–9 seconds — AI analysis. BoardSnap sends the processed image to BoardSnap AI. The model reads every line — printed text, handwriting, diagrams, arrows, lists — and generates a structured summary plus a tri-state action item list. This is where the time is spent, and it varies with connection quality and current API latency.

~10 seconds — Results appear. The summary and action items populate in the app. You can immediately tap to mark items in-progress or done, or start a project AI chat about the board.

## Factors that affect speed

Faster:

• Strong LTE or Wi-Fi connection
• A clean, well-lit photo (less for the AI to work through)
• Smaller boards with less content

Slower:

• Weak or congested cellular connection
• Very dense boards with lots of tiny text, multiple diagrams, and heavy cross-referencing
• AI API peak-load periods (rare but possible)

## Offline behavior

If you're offline when you snap, the capture happens instantly on-device. The AI analysis is queued and runs when you reconnect — so the ten-second clock starts from when the connection is restored, not when you took the photo.

## How does this compare to alternatives?

Manually photographing a board, opening ChatGPT or Gemini, uploading the photo, and writing a prompt to get something structured typically takes 2–5 minutes — and still doesn't produce a tri-state action item list, project memory, or brand-aware output. Ten seconds vs. five minutes is not a marginal difference when you're closing down a meeting room.