Data that shows up already linked.
Upload a receipt, forward an email, snap a check — Muin extracts the fields, links to the right person, detects duplicates, and assigns the right role.
No data entry. No orphan records.
Try the intake patterns running on staging
New-member onboarding (community center), after-school registration (school), volunteer application (disaster response), scholarship application (multi-step). Every intake variant the platform supports, themed for a real use case.
Capture → Extract → Link → Enrich
One pipeline, every channel. Each stage runs automatically the moment new data arrives.
Capture
Upload from the web, forward an email to your tenant address, snap a photo on mobile, import a bank statement, or fire a payment webhook. Every channel lands in the same intake pipeline.
Extract
Eight specialized extractors handle PDF, image, DOCX, legacy DOC, Excel, plaintext, OCR, and AWS Textract. Fields come out with confidence scores. Document types are tenant-configurable.
Link
Invoice vendor auto-links to the existing Organization record. Payer name auto-links to the Person record. Bank row auto-links to a Location + Org. Every link carries a confidence tier.
Enrich
Duplicates detected. Roles assigned. Activity notes auto-generated on the timeline with an "AI Generated" badge. The extracted data is now cross-module, cross-channel, and searchable.
Seven capabilities, one pipeline
Everything that happens between "data arrives" and "data is usable" — automated.
Multi-channel capture
Upload from the web admin, forward to {tenant-slug}@inbound.falaah.ai, snap a photo in Muin Go, import a bank statement, or POST to a webhook. One intake pipeline, any channel.
Document Intelligence →AI field extraction
Eight specialized extractors cover PDF, image, DOCX, legacy DOC, Excel, plaintext, OCR, and AWS Textract. Tenants define their own document types — invoices, W-4s, NDAs, receipts — via DocumentTypeDefinition.
Document Intelligence →Auto-link to unified registry
Invoice vendor → Organization. Payer name → Person. Bank row → Location + Organization. Email sender → Contact. Every link carries a four-tier confidence score.
People Registry →Smart duplicate detection
Fuzzy match on email, phone, and name across Persons and Organizations. Confidence scoring. Dismiss tracking so the same false positive never prompts you twice.
People Registry →Cross-module role assignment
Once a payer identity is resolved, it flows into the role-management system — donor, vendor, employee, contact. Permissions, portals, and workflows kick in automatically.
People Registry →Auto-generated activity notes
Eleven-plus event triggers — first invoice, payment overdue, sentiment drop, donor lapse, case opened, contract renewed, entity merged, health decline — write CRM timeline entries automatically with an "AI Generated" badge and 24-hour dedup. Staff sees the story, not just the data.
Intelligence →Searchable and embedded
Every extracted field and every auto-note is indexed and embedded via pgvector. Ask in plain language — "show me every invoice from Acme" — and get the answer across documents, activities, and payments.
Intelligence →Four real flows, zero data entry
Mailed check
Staff snaps a photo. Check number, amount, and payor are extracted. A DonationIntent is pre-filled, matched against an open pledge, and the tax receipt auto-issues. No typing.
Vendor invoice
Accountant uploads a PDF. Vendor name auto-links to the existing Organization. A payment draft is created and the approval workflow fires for whoever owns that vendor category.
Forwarded email
A donor emails a receipt request. The sender auto-matches to an existing Person record. A response is drafted with the correct donation history attached. Staff reviews and sends.
First-invoice insight
New vendor invoice arrives. Organization is auto-created. An AI-generated activity note — "First invoice received from Acme Corp — $4,200 for Q2 services" — lands on the new vendor timeline with a suggested next action.
Why this matters
Other platforms require manual data entry or leave OCR in a silo. Muin's extraction is the front door to a unified people + org + location registry.
No orphan records. Every extracted field has a confidence-scored link to the right entity, reviewable and correctable at any time.
Works across the entire platform — not just documents, but payments, emails, bank rows, and mobile captures.
Frequently asked questions
Stop typing what you can capture.
Muin's intake pipeline turns every upload, email, photo, and webhook into structured, linked, and searchable data — automatically.
Start free trialBeta Notice: Muin is preparing for beta. Features described on this page may be in development, subject to change, or not yet available. See our beta agreement for details.