Eleven products.One AI runtime.
Every product below shares the same document intelligence and agentic core — configured, not rebuilt, for invoices, purchase orders, ERP conversations and enterprise workflows.
Native and scanned PDF invoices in, structured and validated records out. Layout-aware OCR reconstructs line items even from dense, multi-column tables, then a fine-tuned extraction layer maps every field to your schema — vendor, invoice number, PO match, currency and totals — with a weighted confidence score on each one.
FEATURES
- ✓ Native (text-layer) & scanned OCR ingestion in one pipeline
- ✓ Table-aware line-item reconstruction, no digit/description bleed
- ✓ Multi-currency resolution with weighted evidence scoring
- ✓ Duplicate invoice detection against prior submissions
- ✓ Vendor & PO cross-matching against ERP master data
ARCHITECTURE
TECH STACK
99.2%
Field-level accuracy
4.8s
Avg. processing time
80%↓
Manual keying reduced
Roadmap: multi-language invoice support, inline reviewer correction with active learning.
Purchase orders read, matched and reconciled automatically. PO Intelligence pairs every incoming invoice against its source PO, checks budget codes and approval chains, and flags variance before it reaches your ERP.
FEATURES
- ✓ Three-way match: PO, goods receipt, invoice
- ✓ Budget-code and cost-center validation
- ✓ Approver-chain routing based on delegation of authority
- ✓ Price and quantity variance tolerance thresholds
ARCHITECTURE
TECH STACK
Roadmap: blanket PO support, multi-currency PO reconciliation.
A conversational agent embedded directly in Oracle Fusion, Oracle EBS, PeopleSoft, SAP or Microsoft Dynamics. It answers policy questions, drafts records and posts approved transactions — always inside your existing permissions and approval model.
FEATURES
- ✓ Native panel inside Oracle Fusion, PeopleSoft, SAP & Dynamics
- ✓ Natural-language queries over live ERP data
- ✓ Draft-and-confirm posting, never silent write-back
- ✓ Role-aware — respects existing ERP permissions
ARCHITECTURE
TECH STACK
See the full ERP-by-ERP breakdown on the ERP Modernization page →
Every extracted field is cross-checked against configurable business rules, historical patterns and your knowledge base before it's trusted. Confidence scoring follows a worst-wins model — one failed check sends the record to review, it never averages out.
FEATURES
- ✓ Rules and thresholds live in configuration, not code
- ✓ Weighted evidence scoring per field
- ✓ Worst-wins confidence aggregation
- ✓ Full audit trail of every rule evaluation
ARCHITECTURE
TECH STACK
Ground every model response in your own policies, contracts and ERP reference data — with citations, not guesses. The Knowledge Engine indexes vendor master data, tax rules and internal policy documents so every extraction and answer can point to its source.
FEATURES
- ✓ Cited answers grounded in your source documents
- ✓ Vendor, tax and policy reference indexing
- ✓ Incremental re-indexing as policies change
ARCHITECTURE
Multi-step AI agents that plan, call tools over MCP, and hand off to a human only when it matters. Agents chain document intelligence, validation and ERP actions into a single, auditable workflow.
FEATURES
- ✓ Multi-step planning with tool use over MCP
- ✓ Human-in-the-loop hand-off at configurable checkpoints
- ✓ Never raises silently — every failure produces a reviewable record
TECH STACK
Configurable, staged workflows that route documents and decisions through approvals, exception queues and downstream systems — without a single line of custom orchestration code.
FEATURES
- ✓ Visual, config-driven workflow builder
- ✓ Exception queues with SLA tracking
- ✓ Additive, backward-compatible rule changes
ARCHITECTURE
Don't see your process listed? We build custom pipelines on MetaFlowKit for whatever your enterprise stack needs next — from expense reports to contract intelligence to logistics documents.
See engagement models on the Services pageUNDER THE HOOD
The engines every product shares.
Three specialized modules power every product above — configured per use case, never duplicated.
OCR & Geometry Engine
Layout and column-detection engine that reconstructs tables from scanned documents without digit or description contamination.
Knowledge Engine
Indexes reference data and policy documents, surfacing grounded answers with citations for every claim.
Currency Resolution Engine
Weighted evidence scoring resolves currency and locale ambiguity across global vendor documents.
See the platform on your own documents.
Run the live demo, or talk to our team about a pilot.