Every day I receive 100–120 orders and invoices that arrive in four different guises: PDF files, WhatsApp chats, screenshots or photos, and occasional structured text exports. My goal is to turn this mixed stream into clean, reliable data that lands automatically inside our Silo WMS under the correct Item, Customer, and Vendor records. The workflow I envision is straightforward for the user but smart under the hood: • Ingestion layer that watches an email inbox, a WhatsApp Business API number, and a shared cloud folder for new PDFs, images, or text snippets. • AI-assisted parsing that detects the document type (order vs. invoice), performs OCR where needed, and extracts the key fields—SKU, quantity, price, customer code, vendor reference, dates, etc. • Normalisation & validation against our existing Silo master data so mismatched SKUs or unknown partners are flagged before posting. • Final hand-off to Silo WMS via its API (or flat-file import if you prefer) with a clear audit trail of what was received, parsed, and posted. Deliverables 1. A repeatable, containerised pipeline (Python, Node, or comparable) with source code, README, and environment files. Use of AI tool will be better. Prefer Low to No Code. 2. Configuration for document classifiers/OCR models—Tesseract, AWS Textract, Google Vision, or another engine of your choice—tuned to our layout samples. 3. API or file-based integration scripts that create or update orders and invoices in Silo WMS. 4. A lightweight dashboard or log view so my team can review exceptions and re-queue any failed documents. 5. Test suite and sample data proving the system copes with the stated 100-120 docs per day at >95 % accuracy. Acceptance criteria • End-to-end run on a provided sample set posts all valid orders into Silo with zero manual edits. • Misreads, missing fields, or unmapped masters surface clearly in the exception queue. • Average processing time per document under 30 seconds. If you have previous experience marrying OCR/NLP with a WMS—or have clever ideas about WhatsApp automation—let’s get this flowing.