AI Agent Core
The open-source foundation for Tabularium AI agents. Fork the core, retain only the capabilities required for a workflow, and add its integrations.
Explore Agent CoreDocument Processing
Transforms document images into structured, reviewable record data through OCR, indexing, and verification - establishing the trusted foundation for downstream workflows.
Recognition
OCR engines restore structure from scanned images by removing visual noise and reconstructing the logical flow of content, including zoning headers, footers, and inline noise.
Indexing
Classifies document type, page purpose, and segment role while indexing parties, clauses, references, and key fields for legal searchability and downstream workflows.
Verification
Verifies indexed results with Tabularium AI IQ checks, ambiguity detection, and explainable review signals - enabling targeted reprocessing and higher-confidence output.
Advanced Processing
Applies record-level analysis and system logic to enrich data, calculate outcomes, and analyze related records - supporting readiness, compliance, and decision-making.
Enrichment
Parses and enriches legal descriptions with structured parcel data and spatial logic. Also detects references to exhibits, attachments, and prior recordings - enabling linkage across document sets.
Computations
Calculates statutory fees, fund splits, and county/state allocation logic based on transaction data and preconfigured rules.
Title Analysis
Runs title-style analysis on related records - checking chains, gaps, and conflicting data to surface risk signals before recording or downstream use.
Image Processing
Prepares the visual record for compliant use through redaction, image refinement, and official output generation - ensuring submission-ready and archive-ready deliverables.
Redaction
Automatically detects and redacts protected data using document classification, contextual field understanding, and jurisdictional rules to apply precise, defensible redactions directly to the image across official record workflows.
Visual Refinement & Conversion
AI-driven refinement enhances image clarity, alignment, and visual structure, followed by automated conversion into standardized representations optimized for integration, verification, and archival workflows.
Endorsement & Certification
Generates recording confirmation pages and certifications programmatically - styled per subscription and automatically inserted into output packages.
Security, Privacy, and Compliance
Tabularium AI is designed for high-trust official-record workflows where access control, data isolation, retention, redaction, auditability, and human review all matter before records move downstream.
Authenticated Access
API calls use subscription-specific credentials. Keys can be rotated, scoped to the active subscription, and protected behind HTTPS-enabled infrastructure.
Subscription Isolation
Each subscription separates credentials, private exchange storage, processing context, configuration, usage tracking, and customer-specific rules.
Private AI Processing
AI workloads run in isolated Azure AI Foundry environments with no cross-tenant sharing. Processing environments can support HIPAA-compatible deployment needs.
Data Protection Controls
Controls are organized around document custody: protect access, minimize persistence, redact protected content, and retain evidence for review.
Retention and Purge
Transient artifacts created during document processing and fine-tuning are purged by policy. Permanent customer records remain in the customer's environment.
Protected Data Redaction
Redaction detects protected values using document classification, contextual field understanding, and jurisdiction-specific rules before output packages are released.
Customer-Controlled Storage
Generated metadata, redacted pages, endorsements, and certificates can be delivered into governed storage controlled by the customer.
Governance and Auditability
Review controls keep automation useful without removing accountability from official-record operations.
Compliance Gates
Tabularium AI IQ checks, ambiguity detection, and configurable rules help block weak results before records move into submission, export, or publication workflows.
Audit-Ready Explanations
Indexing, verification, fee logic, and exception handling produce explainable review signals so teams can understand why a result passed, failed, or needs review.
Human Review Routing
Edge cases, policy-sensitive decisions, and low-confidence outputs can be routed to human reviewers while routine checks remain automated.
Capabilities
Official record domains supported across land, court, vital, permit, and certification workflows.
Land & Financial Records
Official records establishing ownership, claims, priorities, and financial
obligations.
Deeds, mortgages, liens, easements, UCCs, tax liens, and bankruptcies - records shaping title, debt, and secured interests.
Vital & Court Records
Recorded documents defining life events, legal status, rights, and court
outcomes.
Births, deaths, marriages, adoptions, name changes, probate filings, judgments, and divorces - records defining identity and enforceable legal status.
Permits & Certificates
Proofs of authorization, compliance, and regulatory approval.
Occupancy, notary, and environmental capabilities - triggering finalization and enforcement.
Coming Q1 2027.
Supported Document & Data Formats
Supported formats for document ingestion, recording, indexing, verification, and integration.
Binary Formats
High-fidelity image formats widely used for archival and legal
admissibility.
- TIFF / TIF (single and multi-page)
- JPEG, PNG (supplemental use)
Document Formats
Digital containers accepted for official recordation and legal workflows.
- PDF (all versions)
- PDF/A-1, PDF/A-2(archival compliance)
- Coming Q1 2027.
Data & Metadata Formats
Structured formats for indexing, integration, and automated verification.
- JSON (index and metadata)
- XML (PRIA eRecording)
Processing Modes
Processing modes available - real-time APIs, scheduled batch, and human-in-the-loop review.
Real-Time API + Webhooks
Transactional API calls process immediately, with results returned inline or through webhooks - ideal for point-of-sale and partner integrations.
Batch (Scheduled)
High-volume, offline jobs run on your schedule - perfect for nightly backlogs or historical migrations.
Coming Q1 2027.
Human-in-the-Loop
Async review queues route edge cases to specialists - meeting strict QA or statutory verification requirements.
Rules Management
The system automatically generates title-class-specific rules and distilled models from public knowledge to drive classification, extraction, fee/fund computation, and downstream behaviors. This removes most of the manual effort required to build class-specific logic while still supporting targeted fine-tuning where needed. These per-class rules are editable and become the source of truth for continuous improvement.
Processing & Classification
Plain-English rules for indexing, enrichment, chaining, confidential handling, and page classification. Tabularium AI compiles them into AI-ready logic - no new syntax to learn.
Weights & Gates (IQ Verification)
Verification formulas for index quality (IQ). Write in plain English; the system compiles them to formulas and can translate them back for review via chat.
Nomenclature
Jurisdiction-specific terms Tabularium AI recognizes to refine indexing, enrichment, chaining, and chat explanations.
Formulas & Markup
Computation logic for fees and allocations, plus layouts, endorsement and confirmation formatting, and visual markup - adjustable per title class.
Title-Class Derivation & Isolation
From a subscription’s masters, the system generates self-contained rules per class - scoped, testable, and safe to evolve.
Runtime Configuration
Derived components remain editable per subscription - tailor behavior and compliance without altering global masters.
Feedback & Fine-Tuning
Submit structured feedback to refine per-class behavior. Once reviewed and accepted, changes are versioned and applied to improve classification, extraction, fees/funds, and explanations.
User-Driven Feedback
Corrections tied to classification, extracted values, fees/funds, or formatting - scoped to the active title class.
Automated Updates
Accepted feedback is versioned and applied to per-class rules - evolving classification, extraction, and computations.
Subscription-Centric Architecture
Subscription defines a secure and isolated execution context - linking API credentials, governed storage, and usage tracking. Organizations can manage multiple subscriptions, each enabling role-based access, billing transparency, and configuration flexibility. Fine-tuning details for configuration are covered in the Fine-Tuning section.
Private Exchange Storage
Each subscription provides a token to private storage that governs access to submitted documents, generated metadata, and binary outputs - such as redacted pages, endorsements, or certificates. All files are purged according to a configurable retention policy.
Dedicated API Key
API calls are authenticated with a subscription-specific key. Keys can be rotated at any time by providing a system-generated source identifier.
Privacy
Our AIs are securely deployed on Azure AI Foundry, with no shared information, in HIPAA-compatible environments. We do not store your data in any storages. All information generated during fine-tuning and document processing is temporary and is properly purged. All permanent data should be stored in your environment - we’re happy to help you set up a secure environment if you don’t have one.
Isolated Azure Deployment
Our AIs run in isolated Azure AI Foundry workspaces with no cross-tenant sharing; processing environments are HIPAA-compatible.
No Data Persistence
We do not store your data. Transient artifacts created during fine-tuning and processing are purged per policy.
Customer-Controlled Storage
Permanent data remains in your environment (cloud or on-prem). We can help you establish a secure, governed deployment if needed.
AI Agents
AI Agent Core is the open-source foundation for Tabularium AI agents. Each specialized agent starts as a fork of the core, retains only the capabilities required for its workflow, and adds the integrations it needs.
REST or MCP Integration
Choose REST today or prepare AI assistants, applications, and agents for MCP connectivity.
REST Integration
Connect existing systems through versioned endpoints, secure authentication, and documented request and response schemas.
MCP Integration
Connect AI assistants, applications, and agents directly to Tabularium AI capabilities through Model Context Protocol.
Coming soon.
Secure & Versioned
Use site-key access, API keys or OAuth 2.0, additive API versions, and clear deprecation timelines.
Applications or Components
Deploy intake and processing applications that connect directly to Tabularium AI workflows.
Tabularium AI Lens
Embeddable TIFF document workspace for source-image review, AI index context, search, selection, and page assembly inside existing systems.
Online Portal and Kiosk Intake
Intake-only self-service applications for remote, lobby, and counter workflows that produce standardized ready-to-record packages with fix-lists.
Coming Q3 2026.
Zero Legacy Integration
Manifest-based delivery appends visual summary pages for clerks while exporting XML indexes for import, letting Tabularium AI fit legacy recording systems without changes.
Billing & Invoicing
Tabularium AI uses precise usage-based billing rules powered by an industry-leading platform. Each invoice reflects real operations - charged per tool per page - without bundling or estimation. All processing, tuning, and refinement actions are accounted for to ensure fairness, flexibility, and full transparency. View the usage-based billing walkthrough
Usage-Based Pricing
- Charged per activated tool per page - no flat fees or bundle-based pricing
- Billable service categories:
- Recognition, Extraction & Enrichment: core interpretations
- Image Processing: redaction, markups, enhancements
- Advanced Processing: computations, composition, title analysis, verifications
Invoicing and Payments
- Invoices are issued periodically based on actual service usage
- Supports multiple secure payment methods and currencies
- Payment methods and terms are configurable per subscription
- Billing is handled through a globally trusted platform with full compliance, auditability, and failover support
Usage Reports
Detailed usage logs are available upon request. Reports include operation-level insights that map directly to billing.