At OpenText World 2026, we opened the Content Cloud showcase with a simple point of view: Agents do not fail on models. They fail on content. Most organizations have already run an AI pilot. When it disappointed, the model was not the problem. The agent could not find, access, understand, or trust the content it needed.
We built the showcase around six things clients told us this year, from making migration easier to meeting people in the business apps they already use. All week, those same asks came up in breakout sessions, demos, and hallway conversations. So instead of recapping the keynote, I asked the product marketing leads across our Content Cloud portfolio one question about what their clients should take away from the week.
Here is what they told me.
OpenText™ Content Management with Tom Grucza, Sr. Product Marketing Manager, Content Management
Tom, your client panel was the most-attended session at OpenText World. What did the clients on stage say that every Content Management client should hear, and what does it tell us about where they are taking their content next?
The biggest lesson from our client panel was that the real value of Content Management comes from bringing content into the business applications people already use, and that this is also what prepares them for AI. Attendees saw peers with real constraints solve problems they recognized from their own work, and that’s what people were still talking about after the session.
- ExxonMobil showed how it brought documents and key business processes together into one experience built on Content Management and connected to its ERP and line-of-business systems. It retired a set of custom applications, brought scanning and email intake into the same platform, and gave users a single home page with search, workflow status, and document preview in one place, significantly reducing clicks in a user’s day.
- Siemens Healthineers focused on governance. It connected submission, review, approval, electronic signature, publishing, and periodic review into one controlled lifecycle. Metadata starts reviews automatically, dashboards show who needs to act, and every decision leaves an audit trail, so being ready for an inspection is part of the daily process. A point that has stuck with me since: the biggest value didn’t come from automating one approval it came from connecting every step.
- Our automotive services client took the integration point furthest. He talked about integration as a way to deliver the right document to the right person at the right time, inside the applications they already know, use every day, and like using, instead of as a back-end IT project. If people have to leave their system to find a file, adoption drops and side copies pile up. If the content comes to them, the governed version becomes the easiest one to use.
The three companies are in different industries and are at different points in their journeys, but they agreed on one thing: Content Management delivers the most value when it’s connected to people, systems, and workflows, governed without slowing users down, and built on a platform they already run. Every Content Management client should take one message from this: The integrations and governance you build today to serve your users are the same ones that will let your agents work tomorrow.
OpenText™ Documentum™ Content Management with Dianne Jolly, Sr. Product Marketing Manager, Documentum
Dianne, what did OpenText World prove a regulated client can do with AI today, without moving controlled content out of the environment it must stay in?
For regulated clients, the biggest takeaway from OpenText World is that you don’t have to choose between control and AI. Two Documentum clients showed that the discipline that keeps them audit-ready is the same foundation that makes AI safe to use. And because they control where their content and their AI run, they can move forward without waiting for the rules to change.
- Southern Nuclear showed what that looks like when operating under export controls that require certain nuclear design data to stay in the United States under tightly controlled access. Their content stays in Documentum on premises, in their own data centers, and only the AI models run in the cloud. That flexibility set OpenText™ Content Aviator™ apart—no copies to manage, a single source of truth, and no custom solution to support. The company’s message to the room: Regulated industries tend to be conservative, but that does not stop them from innovating.
- Southern Nuclear started where staff feel the friction every day—searching licensing documents and procedures in plain language means less time searching and less reliance on subject matter experts. Next up is work that takes more reasoning, like assessing how one change affects everything else, and mapping which documents point to other documents, since missing references can create regulatory risk. They were candid about the hard part, too: Real data is messier than users expect, with scans, duplicates, and incorrect metadata, and bad data shows up as incomplete or generic answers.
- A global life sciences manufacturer showed the other half of the story: getting content under control at scale. In 2021 it had multiple document systems, more than 15 paper-based processes, and no standard approval process across its sites. Today, dozens of sites run on Documentum. Standard templates, folders, and built-in client approvals mean users create only the document types the business needs, and a document cannot be used before the client signs off. Even paper stays governed. With some sites issuing thousands of printed records a month, two-person approvals and saved queries help the team find gaps before an auditor does. Their design principle is worth borrowing: The preferred path must be the easiest to follow.
Put the two stories together and the pattern is clear. Approved versions, consistent metadata, and permissions that hold are what auditors ask for. They are also what an agent needs before you let it summarize a procedure, let alone act on one. If you run Documentum in a regulated environment, start with one high-value use case where your content is already well governed, prove it with a small group, and expand from there.
OpenText™ Core Content Management with Cassandra Tilson, Sr. Product Marketing Manager, Core Content
Cassandra, why does a platform that stays current every quarter matter more in an agentic AI world, and what can partners now build or deliver on Core Content that they could not a year ago?
In an agentic AI world, staying current matters just as much as getting current. For Norfolk Southern, automatic updates were a key reason to choose a SaaS platform. They were managing content in two separate legacy applications and knew they needed to consolidate. Moving to Core Content allowed the IT team to shift their focus from upgrades, patching, and security remediation to supporting the business. But modernizing is only half the equation. Core Content clients do not need to plan major upgrades every time AI technology advances because the most current AI capabilities are automatically available every 90 days.
A platform that stays current also creates new possibilities for partners. Core Content delivers open APIs, no-code and pro-code workflow, and ready-to-use configurations. Now, partners can also use OpenText™ Aviator™ Studio to build and orchestrate AI agents on top of governed enterprise content. This enables partners to deliver solutions that connect content to the applications where clients already work, automate content-intensive processes, and create agents grounded in governed enterprise information.
Contesto offers a concrete example. The OpenText Solution Extension Partner has developed an integration connecting Microsoft Dynamics 365 solutions directly with Core Content. Dynamics 365 supplies the business context, while Core Content provides secure, governed enterprise content. Contesto connects the two so users can access relevant content directly within Dynamics 365, supporting business processes across finance, operations, sales, and customer service.
And integration is only the starting point. In an accounts payable scenario, for example, OpenText™ Core Capture can capture invoices and extract information, Core Content can govern those documents within business workspaces, Contesto can surface them directly in Dynamics 365, and Content Aviator can help users ask questions, summarize information, and identify potential exceptions.
That is the bigger opportunity created by a continuously updated SaaS platform. Partners are no longer just extending content management. They can build intelligent, industry- and process-specific AI content management solutions on top of a foundation that is evolving to keep pace with AI.
A big thank you to all our partners who showcased their innovations with Core Content at OpenText World: Amexio Group, Contesto, Deloitte, and Reva Solutions!
OpenText Content Aviator with Sanjana Nair, Sr. Product Marketing Manager, Content Aviator
Sanjana, what is the shortest path from a client’s first prompt to an agent running a real business process, and what did we show at OpenText World that proves it works?
The shortest path for an agent to do real work starts with pre-built agents on your governed content, and it reaches your people where they already work. The order matters. Before anyone builds, the layer underneath must be trustworthy.
We walked through how a leading energy provider turns incoming regulatory notices into organized workspaces with Content Aviator. The intake work someone used to do by hand—reading the document, creating the workspace, filling in metadata, assigning owners—now happens on arrival. What struck me talking to people afterward wasn’t “Look how much time this saves.” It was the recognition that this routine work is exactly what AI needs to be accurate. Skip it and adoption stalls, because the outputs can’t be trusted.
Then we showed where that pattern goes next. The demo put us inside a major airport authority receiving a Civil Aviation Authority permit for a terminal expansion. The permit arrives as prose; safety, security, and environmental conditions, with deadlines in plain language like “No later than 120 calendar days following operation start.”
Pre-built Content Aviator agents did the heavy lift first. The knowledge graph agent understood the ontology. The workspace agent created a business workspace named for the permit reference, pulled each obligation into its own sub-workspace sorted by subject, filled in metadata, and preserved parent-child links back to the source.
A permits and obligations supervisor agent in Aviator Studio sat on top of that foundation. It was written in plain language—what to watch for, which tools to use, how to name what it created—and the pre-built agents were exposed as MCP tools inside Aviator Studio, so the supervisor could call them directly. The finished agent inherits the client’s permissions and deploys through the platform.
The part the audience leaned into was what came next. For most of these users, Microsoft Copilot is the AI assistant they open first. So, the question isn’t whether Content Aviator can do the work; it’s whether the work shows up where they already are. Content Aviator could pull which obligations were due, who was assigned, and the context behind them. Through the Work IQ MCP connection, the same tools Copilot itself uses, Content Aviator asked Copilot to schedule the meeting with the obligation details already in the description. The user never left Copilot. A human confirmed the final step. Outputs inherited source classification. That’s the moment it clicked for me; the governed content isn’t asking users to come to it, it’s going to where they already are.
Two things from OpenText World made that chain possible. Aviator Studio is now included with Content Aviator so clients get the building environment from day one. And open protocols are live, which is what lets Content Aviator meet users inside Microsoft Copilot, SAP Joule®, Google Gemini, or Salesforce Agentforce instead of asking them to switch tools.
So when I’m asked how we keep this safe while teams build, that’s my answer. The foundation is pre-built and governed. The custom layer sits on top.
OpenText Capture and Intelligent Document Processing with Sabra Goldick, Sr. Product Marketing Manager, Capture, IDP and Information Archiving
Sabra, agents are only as good as the data they receive. Which OpenText information capture update from OpenText World most improves what happens downstream, and for which kind of client?
The capture-related announcement that will do the most for AI agents is OpenText™ Capture Aviator™ because it reads the hardest documents the way a person would and brings a person into the loop when something looks off, so that downstream AI agents start with trusted information.
An AI agent can only settle a claim or pay an invoice if it has accurate information to work with, and that comes from information capture. Most documents already move straight through OpenText™ Capture or Core Capture. The hard ones, like handwritten forms, complex tables, blurry scans, and rare document types, are where rules-based capture struggles and someone ends up manually fixing them. Capture Aviator is the GenAI add-on for both products that handles the documents clients choose, like complex or low-confidence ones, while their proven capture workflows keep running as usual. And because model access comes through OpenText, there’s no separate AI contract to negotiate.
The demo that made this real for me was a handwritten motor vehicle accident report, the kind a claims team sees every day. Capture Aviator did three things. It read the handwriting, so the driver’s statement came out as structured text instead of an image someone has to retype. It compared the written statement with the collision diagram and explained that the street names were swapped. And it didn’t guess which one was right. It flagged the mismatch and routed it to a person, so the exception was caught before the document moved on. AI does the heavy lifting, people make the judgment calls, and the agent gets reliable data it can act on.
Who benefits most? Organizations whose straight-through processing has stalled at the most challenging documents: claims teams with handwritten forms, accounts payable teams with invoices full of line-item tables, and any team stuck keying in poor scans or rarely seen forms by hand. Fix the document at the front door and every agent downstream inherits clean data. Fix it later and mistakes either turn into costly rework or slip through unnoticed while agents keep using them.
OpenText™ Information Archive with Sabra Goldick, Sr. Product Marketing Manager, Capture, IDP, and Information Archiving
Sabra, agents need history, not just current content. How does archiving help organizations retire legacy applications without leaving that history behind, and for which kind of client?
Retiring a legacy application doesn’t have to mean losing what’s inside it, and with Information Archive, those old records become governed information AI agents can use.
One of the client asks we shared at OpenText World was “Help us get to AI without a painful migration,” and archiving is a bigger part of that answer than most people expect. Agents don’t fail on models. They fail on content. A lot of the content organizations need is still sitting in legacy applications. Those applications stay online because someone might need the records for an audit, a legal hold, or a business question. So, the costs keep coming, and the information stays out of reach for AI. That’s why I think of application decommissioning as data supply, not cleanup.
OpenText Information Archive is how that works in practice. When an organization retires an application, its records move into one archive that holds structured and unstructured information from active and retired applications. The records keep the metadata and relationships that give them context. Retention, legal hold, privacy, and audit controls come with them, so the compliance record stays intact. The application and its running costs go away. Then the records go back to work. With the Content Aviator integration, people can search, summarize, and analyze archived information without bringing the old system back online. That means agents work from governed records instead of whatever they can find. This matters most for organizations moving to the cloud, consolidating data after a merger, or working in regulated industries, where records have to stay compliant long after the application is gone. The system retires. The information stays.
What ties it together
Six product areas, one pattern. The OpenText clients making real progress with AI did not start with the model. They started with their content: connected to the applications people use, governed where it must live, captured cleanly at the front door, and kept within reach after old systems retire. Few of them called that work AI readiness, but it is exactly what lets an agent find the right version, understand its context, and act on it responsibly.
Thank you to our clients and partners who shared their work at OpenText World this year. And in case you missed it, you can stream the Content Cloud keynote on demand. See you next year!
Microsoft, Copilot and Dynamics 365 are trademarks of the Microsoft group of companies. SAP and Joule are trademarks of SAP SE. Google and Gemini are trademarks of Google LLC. Salesforce and Agentforce are trademarks of Salesforce, Inc.
