# Read the page like an expert · Documents · Dscvry > How the AI reads documents: layout, tables, forms, handwriting, stamps and signatures, with every value tied to where it sits on the page and checked against ground truth. Source: https://dscvryai.com/technology/sense/documents/ # Read the page like an expert. Every field. Where it sits on the page. Whether it's right. What the AI can read · How it works · Wherever the work is. Hosted Private cloud On-premise Edge Air-gapped ## The data you need is stuck on a page. Bills of lading, certificates, permits, invoices and safety data sheets arrive as scans, photos and PDFs in a hundred layouts. Someone retypes them, or they're never read at all. ### A hundred layouts. Every supplier, carrier and authority formats the same information differently. ### Scans, stamps and handwriting. The parts that matter most are often the hardest to read: the handwritten correction, the stamp, the signature. ### Nobody knows how accurate it is. Extraction tools report confidence, not whether they were actually right. ## What the AI can read. Recover the structure, not just the text. ### Layout & tables Sections, tables, rows and columns rebuilt from the page, so a composition table in a safety data sheet comes out as a table, with every CAS number checked. See it in supplier quality → https://dscvryai.com/use-cases/supplier-quality/ Read what people wrote in the boxes. ### Forms & handwriting Permits, checklists and inspection forms read field by field, including handwriting, so an empty signature box is noticed rather than skipped. See it in safety checks → https://dscvryai.com/use-cases/safety-checks/ Check who signed, stamped and when. ### Stamps & signatures Stamps, signatures and dates located and checked against who is allowed to sign, and when the signing should have happened. See it in supplier quality → https://dscvryai.com/use-cases/supplier-quality/ Know how often it's right. ### Ground truth Every field is scored against documents a person has checked by hand, kept apart from anything used to build the extractor. Fields that don't agree go to a person. See it in invoice errors → https://dscvryai.com/use-cases/invoice-errors/ ## The simplest reader that gets it right. Most of a document can be read by rules and small models. The expensive steps run only on the parts that need them, and nothing is trusted until it is checked. Every page Sense Scans, photos, PDFs and emails, straightened and cleaned. Text & layout Specialized model Text, tables and fields located on the page. Rules & checks Rules & code Check digits, formats, totals and cross-field rules. Hard parts Specialized model Handwriting, stamps and damaged scans read closely. To a person Check → Operate Only fields that disagree or fail a check. ## Two weights on one page. The correction matches. Bill of lading HLSU 5521907 · gross weight. Read, checked, and passed to a person where the page disagrees with itself. **Typed.** Gross weight 18,420 kg, in the printed weight box. **Handwritten.** 18,240 kg written over it and initialled. **Other documents.** The packing list totals 18,240 kg. **Rule.** Weights on the bill of lading and the packing list must agree. Illustrative example. Shipment, documents and values are fictional. ## Built for documents that matter. A value you can't trace back to the page isn't evidence. - ### Every value keeps its place. Each field links back to the exact spot on the page it came from. [How Understand works](https://dscvryai.com/technology/understand/) - ### Measured, not assumed. Accuracy is scored against hand-checked documents, field by field. [How Check works](https://dscvryai.com/technology/check/) - ### Documents stay inside. Runs in your environment, including fully air-gapped. [Security & trust](https://dscvryai.com/security/) ## Where it shows up. Problems that start with a document. Which invoices are we overpaying? ### Invoice errors Catch overbilling, duplicates and mistakes before payment. See how → https://dscvryai.com/use-cases/invoice-errors/ Is the supplier’s evidence complete? ### Supplier quality Find missing or inconsistent evidence. See how → https://dscvryai.com/use-cases/supplier-quality/ Does this PO match what we’re paying? ### Purchase orders Find mismatches before payment. See how → https://dscvryai.com/use-cases/purchase-orders/ What here isn’t safe? ### Safety checks Flag what’s unsafe or incomplete. See how → https://dscvryai.com/use-cases/safety-checks/ ## Where it runs. The AI goes to where the work is, from hosted to air-gapped. ### A managed environment. Start with a hosted service when speed of adoption and managed operations are priorities. Agree data handling, access and service requirements as part of the engagement. Discuss hosted · ## The other capabilities. Each one is tested on your own work before you rely on it. ### Sense It sees, hears and reads the work as it happens. Explore Sense → https://dscvryai.com/technology/sense/ ### Understand From piles of paper to records you can use. Explore Understand → https://dscvryai.com/technology/understand/ ### Check Every line compared. Only the gaps reach you. Explore Check → https://dscvryai.com/technology/check/ ### Investigate Follow the evidence across systems. Find out what happened, and why. Explore Investigate → https://dscvryai.com/technology/investigate/ ### Anticipate Know what comes next, and plan for it. Explore Anticipate → https://dscvryai.com/technology/anticipate/ ### Operate The next step, taken. The reason, recorded. Explore Operate → https://dscvryai.com/technology/operate/ ## Bring us the work where wrong is expensive. Start with the operational problem. We’ll work out the data, models and agents it needs, and where it should run. [Talk to us →](https://dscvryai.com/contact/) **London**Discovery AI Limited · United Kingdom **Hyderabad**ReDscvry Technology Private Limited · India **Chennai**ReDscvry Technology Private Limited · India