Unlock the business context hidden in your documents
See how Reltio AgentFlow Unstructured transforms contracts, forms, and documents into governed, connected data for operations, analytics, and AI.
Contracts. Supplier agreements. Customer onboarding forms. Insurance claims. Product specifications. Emails.
Every enterprise creates thousands of documents containing critical business information, yet much of that information remains locked inside PDFs, Word files, and email attachments. Because operational systems rely primarily on structured, governed data, this unstructured information often sits outside the systems that run the business, leaving valuable context inaccessible and difficult to act on.
Industry estimates from IDC and Gartner consistently put unstructured data at 70 to 90 percent of all enterprise data, meaning most of what an organization knows about its customers, contracts, and products sits outside the systems built to act on it.
Why extraction alone is not enough for unstructured data
Most document AI tools stop once the text has been extracted. But enterprise teams need more than words pulled from a page. They need to understand which customer a contract belongs to, which supplier relationship an agreement affects, which pricing tier applies to a product, and how each piece of information fits into a governed data model.
Extracting text is only the first step. The harder task is turning that information into trusted enterprise data. That requires identifying entities, understanding relationships, preserving source lineage, applying governance, and publishing the results into the systems that run the business.
Introducing Reltio AgentFlow Unstructured
Reltio AgentFlow Unstructured is an AI-powered document ingestion experience built for enterprise master data. Rather than leaving them as isolated files, it automates transforming documents, transcripts, and other unstructured content into governed, graph-based context linked to unified profiles – as part of the Reltio Intelligent Data Graph™. As a result, it allows you to build a comprehensive and trusted system of context for your AI agents, operational systems, and analytical solutions.
The product combines AI-powered entity and relationship extraction, intelligent mapping to your Reltio data model, human verification with confidence scoring, direct publishing into your Reltio tenant, reusable templates for recurring document types, and automated pipelines for large-scale processing.
Figure 1. From document to governed context in the Reltio Intelligent Data Graph
Structured master data provides trusted records for customers, suppliers, products and locations. But much of the business context that gives those records meaning may still live in documents.
AgentFlow Unstructured brings that information into the Reltio Intelligent Data Graph, where it becomes governed, connected enterprise data instead of isolated document content.
Instead of creating another repository of extracted text, each document enriches your knowledge graph with new entities, relationships, attributes and nested attributes, crosswalks, and source lineage under the same governance as structured data.
How it works
The experience follows a simple workflow from upload to governed enterprise data.
Step 1. Upload a representative document
Start with a contract, onboarding form, policy document or any file containing business information. Select the target Reltio tenant where extracted data will be published. If you’ve processed similar documents before, attach an existing template to accelerate extraction.
Figure 2. Upload a document and select the target tenant
Step 2. Review AI extraction
AgentFlow displays the original document alongside the extracted structured output so teams can immediately review how contacts, organizations, products, contracts and other entities were identified and mapped.
Need adjustments? Use plain-language instructions to refine the extraction using ‘Finetune extraction’ rather than editing complex configuration by hand.
Figure 3. Review the document and extracted JSON together
Step 3. Verify mappings before publishing
Before anything enters your tenant, data stewards review entity types, attribute mappings, relationship structures and confidence scores generated for every extracted value.
Nothing is published until it is approved. Human review keeps the process governed without slowing down the workflow.
Figure 4. Verify extracted entities and mappings
Step 4. Publish once – or reuse forever
After verification, teams can publish directly to the Reltio tenant, publish and save the extraction as a reusable template, or save the template for future document processing.
Templates remove repetitive setup. Once a document type has been reviewed and approved, the same logic can be reused across similar documents.
Figure 5. Publish extracted entities or save as a template
Scale from one document to continuous processing
Most document initiatives begin with one file. AgentFlow Unstructured lets organizations start small without rebuilding later.
Once a template has been validated, it can be attached to a recurring pipeline that automatically processes new documents as they arrive. The same extraction logic moves from proof of concept to production.
Pipelines support scheduled ingestion, source crosswalk configuration, Data Change Request (DCR) workflows and automated recurring processing.
Figure 6. Configure a pipeline for recurring document processing
A real example: processing a commercial contract
Consider a four-page B2B commercial contract containing tables, headers and narrative text. In a single extraction, AgentFlow Unstructured identified five entity types and 33 mapped attributes, including contacts, the contract record, organizations, pricing tiers and product groups.
Data stewards verified the proposed mappings using confidence scores before publishing the information into Reltio. The same session also generated a reusable template, ready to process future contracts with similar layouts without rebuilding extraction logic.
The outcome is not simply extracted text. It is trusted, connected enterprise data ready for operational use.
Proof Point 1: multi-year contract tracking in a
data-driven business
A commercial data provider with tens of thousands of customer contracts came to Reltio with a specific problem: they had never systematically tracked the terms buried inside their own contracts. Annual pricing escalators, renewal dates, and contract conditions existed only as text on a page, invisible to their own systems.
Using AgentFlow Unstructured, the company built a single template against their standard contract format and mapped extraction directly into their existing Reltio entity model. No new schema. No transformation layer. What came out of the contract PDF matched their tenant on day one.
The result: contract terms that used to require manual review now become structured, queryable data the moment a document is uploaded. Multi-year pricing terms are visible and trackable at scale, and the same template runs automatically against every new contract as it arrives.
Proof Point 2: unlocking collateral data in banking
A leading regional bank has unified its customer and account domains in Reltio and is now expanding into the product domain. During a recent roadmap discussion, the bank’s team asked a direct question: could Reltio extract structured data from collateral documents, loan files, and appraisals?
The answer is yes. AgentFlow Unstructured already supports PDF ingestion today, and this is exactly the kind of use case it was built for. For a bank, that means loan documents, appraisals, and correspondence become part of the same governed profile as the customer and account record. The collateral tied to a loan is no longer a static file sitting in a document system. It becomes structured, connected data that other agentic processes can act on, whether that’s understanding full customer exposure, flagging duplicate collateral pledges, or surfacing risk before it becomes a loss.
This is the broader promise of AgentFlow Unstructured: a single trusted layer where AI agents don’t just know who your customers are and what products they hold. They can also retrieve context from documents, transcripts, and correspondence that used to be invisible to the platform.
Don’t let valuable data stay trapped in documents
Most enterprises already have the information they need to build richer intelligence about customers, suppliers, and products. Much of it is simply trapped inside documents.
AgentFlow Unstructured helps organizations unlock that information, connect it to existing master data, and continuously enrich the Reltio Intelligent Data Graph with trusted enterprise context.
Rather than treating documents as static archives, organizations can turn them into governed data assets that support analytics, operational workflows, and AI.
AgentFlow Unstructured runs on the same Reltio platform that earned AWS Financial Services Industry Competency status in February 2026. It is designed not as an experimental capability, but for the scale, security, and governance required by regulated, data-intensive organizations.
