checking…
SN
⚡ Agentic Document Digitisation

One engine.
Any industry's documents.

DigiSpec turns verbose, semi-structured documents into structured, system-ready data. It extracts domain-relevant data, not a blind transcription of the source.

The domain-intelligence lives in a business-controlled editable wiki, not in code.
Swap the domain-intelligence wiki while the extraction engine stays the same. Domain documents include CPG Formulations, Food & Beverage Recipes, Engineering BoMs, Invoices, Shipments, Purchase Orders, Freight Documents, Legal documents, and so on.
No code. Domain experts own the rules, in plain English. New industry? Author a glossary, not new software. Formulation is live. The first of many.
How it works — 5 layers, fully automated
🗂️
LAYER 01
Classify
Auto-detects: native PDF, scanned image, DOCX, multi-page TIFF
📄
LAYER 02
Extract
Text parsing for native docs · OCR for scanned images & TIFFs
🤖
LAYER 03
Claude AI
Maps raw text to the Enterprise Logical Data Model — ingredients, specs, refs
LAYER 04
13-Point Check
Confidence engine scores every field. Green / Amber / Red flags for SME review
👩‍🔬
LAYER 05
SME Review
Side-by-side review workspace with source document before ERP approval
🏭
OUTPUT
ERP Ready
Structured data published downstream for ERP, MDM & analytics platforms
Supported inputs & capabilities
📄 PDF 🔍 Scanned Image 📝 DOCX TIFF · JPG · PNG 🤖 Claude AI ✅ 13-Point Confidence Engine 👩‍🔬 SME Review Workspace 🔒 Secured Drives 🏛️ Enterprise Vaults Audit Trail ERP Integration
Documents
Complete
Processing
Errors
DigiSpec v3.0 ClearMindsWork.ai Powered by Claude AI
Dashboard
All extraction jobs, status at a glance
Total📄
Processing
Complete
Errors
Ingredients🧪
Templates📋
3
Document
Source
Status
Items
Confidence
Date
Loading jobs…
New Extraction
Select a template first — extraction is driven by the template schema and only calls Claude AI when you proceed. Use Unknown / New Format only if no template fits.
🧪
Sandbox environment — best used with sample or anonymised documents
DigiSpec is a live demo. It works with any formulation or specification document, but we recommend using sample or anonymised files rather than documents containing real product formulations, supplier details, or commercially sensitive data. This isn't a technical limitation — it's just good practice. Production deployment with full data governance controls is available as a dedicated instance. Get in touch if you're evaluating for enterprise use.
Step 1 — Select Extraction Template
🚧
Template creation is coming soon. The three built-in templates (Tablet, Shampoo, Raw Material) are live and drive extraction today. Custom template authoring — where Claude drafts a template from your description — is being wired up next. For now you can browse the built-in templates below.
Template Management
YAML templates drive the extraction engine. Each template defines which fields to extract and how to validate them.
👤 Public
DIGITISED DATA
SOURCE DOCUMENT
FIELD INSPECTOR
Click any field to inspect
Get Tester Access
We review each application personally · SME access granted within 1 business day
✦ New Template Creator
Claude AI drafts a YAML template from your description, grounded in the DigiSpec LDM and existing templates.