Copilot and agents answer from whatever they can reach, and most organisations cannot say what that is. AI readiness is the work of finding out, deciding what counts, bringing it under records control and connecting it to AI tools on terms you can defend. WyldLynx does that work with OpenText Data Security Platform and Content Manager, and with records practitioners who have been making information answerable since 2009.
For twenty years a badly organised file share was a private embarrassment. An AI assistant turns it into an answer: the superseded policy, the draft that was never approved, the spreadsheet with the card numbers, the board paper in an open folder. The tools respect the permissions they find, and the permissions are the problem. Microsoft's own Copilot guidance opens with oversharing for that reason.
So the readiness questions are information architecture questions before they are technology questions. What do you hold, and where? Who owns it? Which of it is a record, which is sensitive, which is duplicate or obsolete? What must be kept, for how long, and what may an AI tool read, change or create? Organisations that can answer those get safe value from AI quickly. Organisations that cannot spend the first year of the AI programme finding out.
Faster correspondence, fewer avoidable corrections, more time for complex cases. A useful AI initiative begins with a recognisable business outcome and a measure of today's performance. Sometimes clearer guidance or a simpler process delivers the gain before any AI does.
The people doing the work know its exceptions and informal steps. Their role shifts from finding and composing to checking, resolving ambiguity and approving, which needs time, access to sources and the authority to reject an answer. Training and an honest conversation about role change are part of readiness.
Business ownership, privacy, security, records management and compliance belong in the design. An AI response, summary or decision may be a business record; the National Archives says records created using AI are managed under the same obligations as any other. Define what is captured, reviewed, retained and disposed of.
Each step produces something the next one needs, and each one is work WyldLynx does every week for records, long before AI made it urgent. People and process run through all five.
Discovery across file shares, SharePoint, Teams, email and business systems with OpenText Data Security Platform: sensitive content, personal information, duplicates, obsolete material and ownership, reported by collection with named owners. Queensland Museum assessed 500 GB in a 30-day proof of value and found 11% duplicates and 38% obsolete files. Data Security Platform and Proof of Value.
A records decision for every collection: what is a record, what is sensitive, what is redundant, obsolete or trivial, and what the retention authority says. Defensible disposal before an AI indexes it; an information asset register and access review for what stays. Our records practitioners make these decisions with your information owners, not for them. Information security and governance.
What counts goes into OpenText Content Manager, or stays where it is under Content Manager's management, with classification, security levels, retention and an audit trail. Capture from Outlook, Teams and SharePoint so the record is made where the work happens. This is the governed foundation an AI programme can be pointed at. Content Manager and Microsoft 365 integration.
Decide what an AI tool may read, what it may change and what it may create, for each use. Copilot works inside Microsoft's permission model; a custom agent needs its own identity, permitted operations, processing locations, logs and tests with real user roles. Where an agent reaches Content Manager, the records system stays the authority: the connector never decides who may see what. Integration and extension options.
Information changes after the clean-up. Automation flags missing metadata, suggests classification, finds duplicates and routes overdue reviews to owners; people resolve the conflicts. AI outputs that are records are captured with their context. The service gets an owner, monitoring and a review cycle, so readiness is a state you maintain rather than a project you finish.
Who approves an AI answer before it leaves the organisation. Who is accountable when a source changes and every summary built on it is now wrong. What an officer does when the suggested route is wrong. Readiness is as much a decision about roles as about data, and the roadmap says who does what.
Most AI readiness advice stops at policy. These two platforms do the work the policy describes, and WyldLynx runs both.
| Readiness question | OpenText Data Security Platform | OpenText Content Manager |
|---|---|---|
| Where is sensitive or personal information outside the governed systems? | Finds and classifies it across connected repositories, with masking at source where supported | Holds it under security levels and caveats once captured |
| What is duplicate, obsolete or trivial? | Identifies it by collection so owners can approve disposal | Runs the disposal under the retention authority and keeps the destruction record |
| Which collections are approved for an AI use? | Profiles candidate sources: sensitivity, age, duplication, ownership | Provides the approved, current, classified set an assistant can be pointed at |
| How does the AI find the right version? | Flags copies and superseded material in the wider estate | Semantic indexing, search and classification inside the records platform in recent releases |
| How does an agent file what it produces? | Not its job | Capture through supported integrations and APIs, with classification and retention applied |
| How do we prove what the AI read and did? | Discovery reports and risk findings over time | Audit trail on every record and access, which survives the record itself |
Neither product makes an AI answer accurate by itself. Together they make the information the answer comes from known, governed and defensible, which is the part of accuracy you control.
Microsoft 365 Copilot, OpenText Aviator, packaged tools and custom development offer different paths. Compare them against your workflows, information sources, controls and full operating cost once steps one to three have told you what you actually hold. Confirm the exact product, deployment and licensing rather than assuming a capability is there.
Distinguish finding an answer from taking action. Search locates information. AI-assisted retrieval and drafting interpret it. An agent chooses steps, uses tools and requests changes across systems. Each carries a different responsibility: finding approved guidance, preparing a response and sending it are three decisions, not one. Specify what the system can access, what it can change, when approval is required and how work continues if it fails. Start with a manageable scope and increase autonomy when the evidence supports it.
Interfaces such as Model Context Protocol let an agent use Content Manager and other systems as tools. The design rule does not change: identity, permissions, approvals and logging are designed and tested, and the records system remains the authority on who may see what.
Current state of the information estate, the people and the controls, against the use cases you have in mind. Where the sensitive, duplicate and ungoverned material is likely to be, what the recordkeeping obligations require, and what Copilot or an agent would reach today.
Data Security Platform applied to a defined sample of your repositories, with an executive report: sensitive content, duplication, obsolete material, candidate collections for an AI pilot and the records decisions that follow. How a proof of value runs.
Use cases ranked by value and readiness, the information work each one needs, Copilot versus packaged versus custom, the controls and roles required, and a sequenced plan with owners. The output is a roadmap your executive can fund in stages.
Capture and integration configured in Content Manager, disposal run, security and ownership fixed, then a pilot on an approved collection with real user roles, measured against the outcome you chose at the start. Microsoft 365 integration and applications built on Content Manager.
Bring the outcome you want to improve and the repositories you are worried about. We will tell you which of the four fits first.
Records practitioners who know what the retention authorities and the archives expect, alongside Content Manager and Data Security Platform engineers who configure it. AI readiness needs both in the same room.
Discovery, defensible disposal, migration, classification and capture, for government agencies, councils, universities and health services, long before AI made them urgent. Queensland Museum is one published example.
Our documents, support requests, work tracking and training run in Content Manager. OpenText published the story. How we use Content Manager.
An OpenText Gold Partner, certified to ISO 27001, with support plans that publish response times. Readiness work hands over to support and training rather than ending at a report. ISO 27001.
Partly, and that part comes last. The readiness that determines whether an AI programme is safe and useful is the state of the information it will read: what you hold, who owns it, how it is classified and secured, and what must be kept. Technology choice follows from that, which is why our approach starts with discovery and records decisions.
No. Start with one outcome and the collections it depends on. Discovery tells you what is in them, records decisions clear what should not be there, and the pilot runs on an approved set. The rest of the estate is sequenced on the roadmap by value and risk.
It makes the information Copilot relies on known, classified, secured and current, which is the part of safety you control. Copilot still answers from everything a user can reach in Microsoft 365, so the permissions and the ungoverned material outside Content Manager need the same attention. How Content Manager and Microsoft 365 work together.
Sensitive and personal information, duplicates, obsolete and trivial material, ownership and age, across file shares, SharePoint, email and business systems, reported by collection. Depending on the components licensed it can also mask sensitive values at source, place holds and carry out approved disposal. Data Security Platform.
Some are, and the National Archives guidance on records created using AI treats them under the same obligations as any other record. Define which outputs are records, capture them with their context and retention, and keep the audit of what the AI read. Read the National Archives guidance.
Contact WyldLynx with the outcome you want to improve and the repositories you are least sure about. We will come back with a starting scope: a readiness review, a discovery proof of value or a roadmap workshop, whichever fits first.
Tell us the outcome you want to improve and the repositories you are least sure about. We will tell you where the sensitive, duplicate and ungoverned information is likely to be, what the records obligations require, and the first step that fits. We take your business personally.
