Falaah Falaah AI

Check Capture on Muin Go

Photograph checks on Android or iPhone at intake — auto-extract check number, amount, payor, and memo; feed the reconciliation queue.

Check Capture turns any staff phone into a check intake scanner. Photograph a check as soon as it arrives — front and back — and the OCR pipeline extracts check number, amount, payor name, memo, and date. The extracted record flows into the Reconciliation queue; the physical check can be deposited on the standard schedule.

When to use check capture

  • At Friday prayers or Sunday service when envelopes are opened the same day
  • At grant-check receipt when the check needs to be attached to the grant record
  • At event cash-and-check intake
  • At a mailbox-sorting session when checks arrive in the mail
  • Any time a donor hands a check to staff before the bank run

The point is to decouple donor attribution from deposit timing. The check can sit in the vault for three days; the attribution is already on the donor’s timeline.

Setup

Device requirements

  • Android: Muin Go app, Android 8.0 (Oreo) or later, rear camera with autofocus
  • iOS: Muin Go app, iOS 14 or later, rear camera

Install Muin Go from the Play Store / App Store, sign in with your staff credentials, and ensure your account has the Capture permission (default for staff in the nonprofits module).

Optional: external scanner

For high-volume intake (100+ checks in a session), a document scanner with auto-feed (Brother ADS-2700W, ScanSnap iX1400) can batch-scan directly into Muin via the web admin. The mobile flow is optimized for 1–50 checks at a time.

The capture flow

  1. Open Muin Go → Capture → Check
  2. Place the check on a dark, flat surface under even light
  3. Tap the camera; the app auto-detects the check rectangle and snaps when the image is in focus
  4. Flip the check over; tap again for the endorsement side
  5. Review the OCR preview: check number, amount, payor, memo, date
  6. If extraction is correct, tap Save
  7. If any field is wrong, edit inline; tap Save

Each capture takes 10–20 seconds. OCR is on-device for amount + check number (fast, offline-capable); payor + memo go to server-side Textract for handwriting support (requires connectivity).

Photo quality tips

  • Lighting: even, diffuse light. Overhead office fluorescent is great. Direct sun creates glare; use shade.
  • Surface: dark background (dark blue, dark gray, black) contrasts against the check and helps auto-crop.
  • Angle: hold the phone directly above the check, parallel to the surface. A slight tilt is fine; extreme angles break OCR.
  • Stability: rest your elbows; auto-snap waits for focus.
  • Handwriting: printed checks OCR near-perfectly. Handwritten amounts and memos are probabilistic — always review the preview.
  • Endorsement: capture the back side too. Muin uses the endorsement to confirm the check has not been previously deposited.

Manual-entry fallback

When OCR fails (damaged check, unusual format, poor lighting), tap Skip OCR on the capture screen and enter the fields manually:

  • Check number
  • Amount
  • Payor name
  • Memo (optional)
  • Date written

The manual-entry flow takes about 45 seconds per check. Photos are still attached for audit.

Offline behavior

  • Capture: works offline. On-device OCR extracts the amount and check number locally; server-side OCR runs when connectivity returns.
  • Queue: captured records are stored locally for up to 24 hours; they appear greyed-out in the Reconciliation queue as “pending sync.”
  • Sync: when the device is back online, captures auto-upload and the queue refreshes.

Offline mode is critical for venues with spotty Wi-Fi (basement prayer halls, remote community centers, pop-up events). Staff can capture a stack of checks without connectivity and sync later.

After capture — where the check goes

Each captured check lands in the Reconciliation queue as a row with its extracted fields. It is matched against:

  • Open pledges (does this check match a pledge the payor committed to?)
  • Existing donor records (does the payor name match a Person in the registry?)
  • Recent bank deposits (was this check already deposited? if so, confirm the deposit reconciles)

When the physical check is deposited on the next bank run, Muin auto-correlates the deposit row (from statement import) with the captured check — flagging any captured check that is missing from the deposit (sitting in someone’s desk) or any deposit row that has no matching captured check.

Common workflows

Friday prayer collection

  1. After jumu’ah, envelopes are opened in the counting room
  2. Two staff members count cash (entered as a cash record with two-person verification)
  3. Checks are photographed one by one on a staff tablet running Muin Go
  4. Bank run happens on Monday; deposit reconciles against the captured checks

Total time: about 15 minutes for 40 envelopes.

Mailbox sort

  1. Weekly, a staff member opens mailed checks
  2. Each check is photographed as it’s opened
  3. The stack goes in the deposit envelope
  4. Deposit happens mid-week; reconciliation closes the week’s captures

Grant check receipt

  1. Grant check arrives via mail or courier
  2. Capture it immediately in Muin Go
  3. The capture’s Memo field links to the grant record (use the grant’s memo code or attach the grant directly from the capture screen)
  4. Deposit the check; reconciliation confirms the grant disbursement row was satisfied by this specific check

Troubleshooting

Issue Cause Resolution
Auto-crop failing Check blends into surface background Use a darker surface; reposition for contrast
OCR misreads amount Handwritten amount, glare, or smudge Edit inline in the preview; submit; OCR training improves
Photo uploads hang Connectivity issue Captures queue locally; sync automatically when online
Duplicate check captured Same check photographed twice Muin detects duplicate check number + amount and prompts to dismiss the second capture
Check doesn’t reconcile with deposit Check was captured but not deposited, or deposited but not captured Open the reconciliation queue; each side flags the missing counterpart