Artificial intelligence can only be as good as the data that fuels it, and most enterprises have an estimated 80 % of their information in unstructured form—documents, drawings, PDFs, images—locked away from the SAP processes that drive the business. In our recent post, “From 80 % Unstructured Data to AI-Driven Insights,” we showed how OpenText Extended ECM (xECM) and SAP together give AI the context it needs to perform. Today we go one layer deeper, exploring how to organize that unruly content so AI can finally work its magic.
AI thrives on quality data, but file shares, SharePoint sites, and silo’d applications keep business knowledge everywhere except where SAP—and, therefore, AI—can see it.
The result is familiar: projects hallucinate, dashboards disagree, and stakeholders question every recommendation.
SAP’s recent event (“SAP’s AI Strategy Relies Heavily on Unstructured Data”) echoed the same point: agentic AI needs governed, contextual data. Yet when your documents are scattered across decades of repositories, how can you deliver that context?
Decades-old repositories hold critical information with zero searchable metadata, and scanned PDFs stored as images remain unreadable to machines.
Many teams hope the AI will “figure it out,” launching pilots with no metadata or governance, while duplicate documents proliferate until no one is sure which version is authoritative. Individually these missteps seem minor; together they cripple AI’s ability to understand relationships and produce reliable insights.
When data scientists complain about “dirty data,” they mean unmanaged documents that aren’t anchored to the SAP records that give them meaning. Stalled investments follow.
Gartner finds that up to 85 % of AI pilots never reach production, often for this reason. Automation savings vanish when an accounts-payable bot can’t match invoices to purchase orders, compliance teams risk fines by serving outdated files, and user trust erodes after a single hallucinated answer.
Qellus fully integrates OpenText xECM inside SAP so every document inherits the same master data—company code, vendor ID, material, asset—that lives in its business transaction. Content is captured and version-controlled in the flow of everyday work: drop an invoice, contract, drawing, or work order into the familiar SAP screen, and xECM instantly applies industrial-grade OCR, auto-classification, and retention rules. The result is an authoritative record that is always current, audit-ready, and governed just like SAP’s structured data.
Because every document now carries rich, trusted metadata, it becomes gold for AI models and agents. The same API-friendly layer streams these annotated files to SAP Joule or any LLM right away—no scheduled batch exports and no need to copy the documents into a separate data lake. Whether you’re matching invoice line items to purchase orders, summarizing contract risk, or guiding field engineers with the exact procedure for a specific asset, the AI sees a curated, context-rich source and can trigger workflows the moment an anomaly appears.
Consider three scenarios.
First, a manufacturer’s 30,000 supplier contracts once sat on a shared drive, searchable only by file name. After ingestion into xECM, each contract is linked to its SAP vendor record, and a Joule Agent surfaces agreements nearing expiry or ripe for renegotiation.
Second, field engineers who once scrolled through 200-page maintenance manuals now retrieve the exact procedure for a specific asset and revision in seconds, even via voice.
Third, audit teams that used to scramble for Sarbanes-Oxley evidence now grant regulators real-time access to a controlled workspace where every policy is versioned and cross-referenced with SAP controls.
Stories like these echo the benefits Henkel and Bosch report with SAP Business Data Cloud and Joule Agents—examples we covered in “SAP’s AI Strategy Relies Heavily on Unstructured Data.”
AI is moving fast, and competitors are not waiting. Without structured, contextual content, even the best algorithms falter; with it, they deliver measurable savings and strategic insight. Qellus brings order to chaos, turning decades of documents into an AI-ready asset anchored in SAP.
Let Qellus guide the journey—from data extraction and metadata design to xECM rollout and AI pilot.
Contact our team to schedule an AI Readiness Assessment.