Generative AI is changing how learning teams begin the content-creation process.

Instead of starting with a blank page, authors can use AI to produce an initial text draft, explore different ways of explaining a subject, or create an image for a learning resource. This can make the early stages of production faster and help teams develop ideas more efficiently.

However, generating a draft is only one part of creating reliable learning content.

Before material reaches a learner, it still needs to be checked for accuracy, aligned with learning objectives, reviewed by the appropriate people, organized into a consistent structure, versioned, published, and maintained.

This distinction is particularly important in 2026.

The central question is no longer simply whether organizations should use AI. It is how they can use AI for appropriate tasks while maintaining control over the quality and lifecycle of their learning content.

A Learning Content Management System, or LCMS, provides the environment in which that control can be maintained.

The Realistic Role of AI in eLearning Authoring

AI can be extremely useful during the first stages of content creation.

For example, an author might use it to:

• Produce a first draft from a clearly defined prompt.
• Suggest headings or alternative explanations.
• Rewrite a passage in a simpler or more concise form.
• Generate ideas for an introduction or learning scenario.
• Create an image to support a lesson or document.
• Develop variations of existing text for further review.

These uses can reduce the time spent on repetitive drafting and help authors move more quickly from an initial idea to editable content.

Nevertheless, an AI-generated response should still be treated as a draft.

AI does not automatically know which organizational procedure is current, which terminology has been approved, whether an explanation is suitable for a particular learner group, or whether a generated claim is factually correct.

Fluent writing can still contain errors, omissions, or information that is inappropriate for the intended context.

Human authors and reviewers therefore remain responsible for:

• Selecting authoritative source material.
• Checking factual and technical accuracy.
• Deciding whether the content supports the learning objective.
• Confirming that the language is suitable for the audience.
• Reviewing copyright, privacy, accessibility, and confidentiality concerns.
• Approving the final material for publication.

This is not a weakness in the use of AI. It is an effective division of responsibilities.

AI can support defined creative tasks. People remain responsible for knowledge, context, judgment, and approval.

Why an LCMS Still Matters

Generating a paragraph or image does not create a complete learning-content operation.

The generated material needs to become part of a managed process. Teams need to know where it belongs, who is responsible for reviewing it, which version is approved, where it has been reused, and how it will be delivered.

An LCMS is designed around this wider content lifecycle.

Unlike an LMS, which is generally focused on learners, enrolments, assignments, progress, and reporting, an LCMS focuses on the creation, organization, reuse, revision, delivery, and maintenance of learning content.

The eXact LCMS supports activities such as project management, content storage, collaborative authoring, reuse, multiple outputs, and delivery through standards-based learning environments.

These capabilities are valuable whether the first draft was written manually or produced with AI assistance.

The LCMS does not need to make every activity AI-powered. Its purpose is to provide structure and control around the content being created.

AI Generation and Content Governance Are Different Functions

It is important not to confuse AI generation with content governance.

AI generation may produce a useful piece of text or an image. Content governance determines what happens to that material afterward.

Governance includes questions such as:

• Has the information been checked against an approved source?
• Is the content appropriate for the intended audience?
• Has a subject-matter expert reviewed it?
• Is this the current approved version?
• Can it be reused in another course or document?
• Who is responsible for future updates?
• In which formats should it be published?

These are primarily content-management decisions rather than generative-AI tasks.

An LCMS provides the tools and processes through which authors, reviewers, project managers, and other specialists can answer these questions.

A Practical Workflow for AI-Assisted Authoring

Organizations do not need to choose between the speed of AI and the control of an LCMS. The two can support different stages of the same production process.

1. Define the Learning Requirement

Before generating anything, the author should identify:

• The target audience.
• The learning objective.
• The required format.
• The authoritative source material.
• The level of risk associated with inaccurate information.
• The people who need to review the content.

A general internal communication may require a different process from technical, safety-related, or regulatory training.

Defining these requirements first helps the author use AI for a focused purpose rather than asking it to create an entire course without sufficient direction.

2. Select the Authoritative Sources

The organization’s approved manuals, policies, procedures, product information, and other controlled materials should remain the source of truth.

The author can use those materials to construct an accurate prompt and to evaluate the resulting text.

The important point is that the AI-generated draft does not replace the original source. The source remains authoritative, and the author is responsible for comparing the draft with it.

This avoids implying that an AI system has automatically verified or understood the organization’s documentation.

3. Generate an Initial Text or Image

The author can then use AI for a specific task.

For example:

Create a concise introduction explaining why employees must follow this procedure. Use a professional tone and do not introduce information that is not included in the supplied notes.

Alternatively, the author may generate an image intended to illustrate a concept, introduce a lesson, or make a learning resource more engaging.

A specific prompt usually produces a more useful starting point than a broad request to “create a course.”

The output should then move into the normal authoring process as an editable draft.

4. Review the Output

The generated content should be reviewed according to its purpose and risk level.

A subject-matter expert may need to confirm technical accuracy. An instructional designer may assess clarity, sequence, and relevance to the learning objective. Other reviewers may need to consider accessibility, brand language, compliance, or localization.

Reviewers should pay particular attention to:

• Unsupported factual claims.
• Outdated information.
• Inconsistent terminology.
• Overly generic explanations.
• Unnecessary repetition.
• Cultural assumptions.
• Images that inaccurately represent the subject.
• Text or visuals that may create copyright, privacy, or confidentiality concerns.

AI can accelerate drafting, but it does not remove the need for professional judgment.

5. Structure the Content in the Authoring Environment

Once the draft has been reviewed, it can be placed into the appropriate templates, layouts, pages, and learning components.

Giotto provides a WYSIWYG authoring environment, customizable templates, responsive layouts, reusable content chunks, interactive elements, and multiple output options. These are established authoring capabilities and should be distinguished from its AI functionality.

This stage turns the initial text and images into a structured learning experience.

The instructional value does not come only from the generated content. It also comes from how the material is organized, presented, connected to activities, and adapted to the learner’s context.

6. Collaborate and Manage Versions

Learning content is rarely created by one person working in isolation.

Authors may need input from subject-matter experts, designers, project managers, compliance teams, customers, or localization specialists.

Giotto supports collaborative authoring through functions such as shared documents, simultaneous work, comments, revision tasks, change tracking, and version management.

These functions help the team distinguish between:

• An initial AI-assisted draft.
• A draft being reviewed.
• A revised version.
• An approved version.
• A previously published version that has since been replaced.

This is an example of the LCMS providing governance around AI-created material without implying that the governance itself is performed autonomously by AI.

7. Reuse Approved Content Carefully

Generative AI makes it easy to create several slightly different explanations of the same subject.

That convenience can also create unnecessary duplication.

For example, different authors might generate separate descriptions of the same safety procedure for a course, manual, presentation, and job aid. When the procedure changes, every variation must be located, reviewed, and updated.

An LCMS offers another approach: once the core information has been reviewed and approved, it can be stored as a reusable content component.

Giotto’s authoring environment is integrated with a shared repository. Authors can search for existing assets, templates, and content chunks and reuse them in other documents. The reuse panel can also help teams evaluate where a changed component is being used.

AI can help create the initial material. Reuse functionality helps prevent the organization from generating a new and potentially inconsistent version every time that material is needed.

8. Publish in the Required Format

After review and revision, the content can be prepared for its intended delivery channel.

A learning object may need to be published as SCORM content, while related material may also be required as a PDF, technical document, or web-based resource.

Giotto supports content export in formats ranging from SCORM to PDF, while the wider eXact LCMS supports single-source content production and standards-based delivery.

This allows organizations to separate content creation from delivery.

The approved content remains a managed asset even when it is distributed through different systems or formats.

Why Responsible AI Use Is Especially Relevant in 2026

In 2026, responsible AI use is becoming more operational.

Organizations are moving beyond informal experimentation and developing clearer policies covering:

• Which AI tools employees may use.
• What information may be entered into them.
• How generated material should be reviewed.
• Whether AI use should be recorded.
• Who remains responsible for the final content.
• When learners or other audiences should be informed about AI involvement.

The European Union’s AI literacy requirements have applied since February 2, 2025. They require providers and deployers of AI systems to take measures to ensure that staff and others using AI on their behalf have an appropriate level of AI literacy.

Certain AI Act transparency obligations also apply from August 2, 2026. Their application depends on the type of system, content, and use case, so this does not mean that every AI-assisted learning asset automatically requires the same form of public labeling. It does reinforce the importance of knowing where and how AI is being used.

Education-technology guidance is moving in the same direction. The 1EdTech Generative AI Data Rubric encourages organizations to examine matters such as AI disclosure, third-party involvement, data use, privacy, user options, and the quality and ownership of data.

For learning teams, AI literacy should therefore include more than learning how to write an effective prompt.

Authors also need to understand:

• The types of mistakes generative AI can make.
• Which information should not be entered into an AI service.
• How to verify generated content.
• How AI output should be documented and reviewed.
• When expert or legal advice may be necessary.
• Why the author remains accountable for the published result.

What Giotto’s AI Capabilities Currently Cover

Clear product communication is particularly important when discussing AI.

At present, Giotto’s AI-assisted capabilities focus on two areas:

1. Text and Image generation
2. Course generation

These capabilities can help authors develop initial written and visual material more efficiently.

The wider functions discussed in this article, including collaboration, comments, version management, content reuse, repository integration, templates, responsive design, and multiple outputs, are LCMS and authoring capabilities. They should not be presented as autonomous AI functions.

Giotto does not need to automate the entire content lifecycle with AI to provide value.

Its current value comes from combining focused AI assistance with an established authoring and content-management environment:

• AI helps the author create text, images and courses.
• Authors and reviewers check the content.
• Giotto supports structure, collaboration, reuse, and version management.
• The LCMS supports the wider content lifecycle.
• People retain responsibility for approval and publication.

This positioning is both realistic and valuable.

It avoids treating AI as a replacement for instructional design, subject-matter expertise, or content governance.

Preparing for Future AI Capabilities

AI-assisted authoring will continue to evolve.

Over time, additional forms of AI support may become useful across the content lifecycle. However, specific capabilities should only be communicated when they are available and ready for customers.

The strongest foundation for future development is not a list of speculative features. It is a well-managed content environment.

When content is structured, versioned, reusable, and connected to established workflows, organizations are better positioned to introduce new forms of assistance responsibly.

As Giotto evolves, the underlying principle should remain consistent:

AI can support clearly defined tasks, while people retain responsibility for accuracy, instructional quality, and publication.

This approach allows innovation to continue without creating unrealistic expectations about what the product does today.

Questions Organizations Should Ask

Before introducing AI-assisted authoring, learning teams should consider a small number of practical questions:

1. Which types of text and images may be generated with AI?
2. What information may authors enter into the AI tool?
3. Which sources must be used to verify the generated material?
4. Who reviews the content before it is published?
5. How will drafts and approved versions be distinguished?
6. How will approved content be stored and reused?
7. Are authors trained to recognize inaccurate or inappropriate output?
8. How will privacy, confidentiality, and intellectual-property concerns be managed?
9. Does the final content meet accessibility and organizational quality requirements?
10. Who remains accountable for the published learning resource?

These questions establish a practical governance framework without assuming that the AI itself performs verification, approval, or compliance management.

Conclusion: Use AI for Creation, an LCMS for Control

The success of AI in eLearning should not be measured only by how much content it can generate.

It should also be measured by whether the resulting material is accurate, useful, maintainable, reusable, and appropriate for learners.

Giotto’s current AI capabilities help authors generate text and images. The surrounding authoring and LCMS environment helps teams structure, review, collaborate on, version, reuse, publish, and maintain that content.

This division of responsibilities is important.

AI can help overcome the blank page and accelerate the creation of initial assets. It does not replace authoritative sources, instructional expertise, human review, or organizational accountability.

In 2026, the most effective strategy is therefore not to automate every part of content production. It is to use AI where it provides clear value while keeping people and controlled content processes at the center.

---

FAQ

What can Giotto currently generate with AI?

Giotto’s current AI capabilities focus on generating text and images, and creating whole courses just using prompts. These functions support authors during content creation but do not replace the wider review and content-management process.

Does Giotto automatically verify AI-generated text?

No. AI-generated text should be reviewed by the author and, where appropriate, by subject-matter experts or other qualified reviewers. Approved organizational documents should remain the authoritative source.

Can AI-generated content be reused in other documents?

Once content has been reviewed and approved, it can be managed as a reusable content component using Giotto and the shared repository. Reuse should take place after review so that unverified drafts are not distributed across multiple learning resources.

Will Giotto include more AI capabilities in the future?

eXact intends to continue developing and exploring AI-assisted functionality, not just on Giotto, but on the whole LCMS environment. This includes the Digital Repository, the Project Manager, and the Translation Module.