Maybe it is 80 slides. Maybe it is 100. Maybe the content is still useful, but the experience feels dated, overloaded, and harder to fix than it should be. The visuals are tired. The text is dense. The flow reflects how someone used to present the material live five years ago, not how people learn best now.

And that is usually the real problem.

It is not that the training has no value. It is that refreshing it manually can take so much effort that the work starts to feel bigger than the payoff. That is where an AI tool like Gamma can help.

Of course, Gamma is not a replacement for instructional design practices. It is a faster way to take legacy content, restructure it, modernize the presentation, and create a stronger draft to work from. For learning teams sitting on old PowerPoints, PDFs, and internal decks, that can be a very useful shift.

How do you refresh old training content effectively?

A lot of older training content still lives in PowerPoint. That makes sense. PowerPoint has been the default home for onboarding sessions, workshops, compliance reviews, product walkthroughs, and internal knowledge transfer for years.

But over time, those decks tend to grow in all the wrong ways. They become longer, denser, and harder to maintain. Important ideas get buried. Every update creates another version. And eventually the deck becomes something everyone knows needs a refresh, but nobody is excited to own.

For instructional designers, that creates a familiar tension:

• The content may still be valuable
• The learner experience is no longer strong
• A full redesign would take real time and budget
• Manually cleaning up every slide often feels like low-value production work

This is where Gamma fits well. It gives you a way to turn an old deck into a cleaner, more modern starting point, so you can spend less time rebuilding slides and more time improving the learning experience itself.

Why Gamma is useful for training teams

What makes Gamma helpful is that it sits in the middle ground.

You do not have to leave the old training deck untouched, and you do not have to rebuild the entire thing from a blank file either. You can import what already exists, let Gamma restructure and restyle it, and then step back in as the designer to refine what matters most.

That matters because the higher-value instructional design work is usually not slide cleanup. It is work like:

• Clarifying learning outcomes
• Reducing cognitive overload
• Improving flow and chunking
• Removing unnecessary content
• Deciding what belongs in slides versus narration, activities, or job aids

If a tool can reduce the production drag, that gives the designer more room to focus on clarity, usefulness, and learner impact.

How to modernize an old 100-slide deck with Gamma AI?

Imagine you have a training presentation that still contains good information, but it clearly shows its age. It was built for a live presenter. The wording is too heavy. The visuals are inconsistent. There is too much content on too many slides. Nobody wants to present it as-is, but recreating everything from scratch feels excessive.

That is a strong use case for Gamma.

Instead of manually redesigning slide 1 through slide 100, you can import the existing material, choose how Gamma should treat the content, apply a more current visual direction, and generate a new version much faster.

It is not the finished product on its own. It is a much better first draft.

Step 1: Import the old file

From Gamma, click Create new and choose Import file or URL.

This is the feature that makes the workflow especially useful for instructional designers. You are not starting from a prompt and hoping for the best. You are bringing in existing material that already contains subject matter, structure, and context.

Gamma supports common formats like PowerPoint, PDFs, and documents, which makes it practical for the kinds of legacy assets learning teams usually inherit.

Step 2: Choose how Gamma should handle the content

After uploading the file, Gamma gives you different ways to work with it. This is where the tool becomes more than a simple visual refresh.

Transform Content

Choose this when the original deck needs more than a facelift. This is usually the better option when the slides are overcrowded, repetitive, or not structured in a way that supports learning well.

Visual Import

Choose this when the original deck is already fairly solid and you mainly want Gamma to improve the layout and presentation style while preserving more of the original structure.

For older training decks, Transform Content is often the more useful choice because it creates room for actual modernization, not just surface-level cleanup.

Step 3: Review the structure before generating

One helpful part of the Gamma workflow is that it does not force you straight into a final output. You get the chance to review how the imported content is being structured first.

From an instructional design perspective, this matters a lot. Structure is often where old training content breaks down. The content may be accurate, but it is not chunked well, it does not flow cleanly, and it asks learners to process too much at once.

This step gives you a chance to pause and think like a designer again:

• What is essential here?
• What can be trimmed?
• What should be regrouped?
• What probably should not be a slide at all?

Step 4: Decide whether to preserve or condense the text

Gamma lets you choose how much of the original wording should carry into the generated presentation.

• Preserve keeps more of the original text intact
• Condense tightens the content into something more presentation-friendly

This is not just a formatting decision. It is a learning decision.

Many older training decks say too much on every slide. They were often built around the presenter’s script instead of the learner’s experience. Choosing to condense can help reduce cognitive overload and surface the real message more clearly.

If the wording is sensitive, technical, or tightly regulated, preserve may make more sense. If the content is bloated and hard to scan, condense is usually the better path.

Step 5: Apply a cleaner visual direction

Once the content settings are in place, Gamma lets you choose themes, language, and output style.

This is where the old deck starts to feel less like a forgotten internal file and more like something learners can actually engage with. And that matters more than it might seem. A cleaner visual system can improve hierarchy, readability, and attention. It can also make the content feel more credible and more current.

For learning designers, the goal is not just to make it look nice. The goal is to make it easier to understand.

Step 6: Generate supporting visuals if needed

If the original deck is visually weak, Gamma also gives you the option to generate AI visuals and graphics. That can be helpful when older slides are mostly text, rely on generic stock imagery, or have no visual consistency at all.

You can choose models and styles depending on the tone you want. In practice, this can speed up the refresh process and help create a more cohesive draft without sending you off into a separate design workflow right away.

Step 7: Generate the draft, then do the real design work

Once you are ready, Gamma generates the updated presentation. This is the point where you move from legacy content to a cleaner draft that is actually workable.

But this is also where the instructional designer needs to step back in fully.

The generated deck still needs judgment. You still need to look at it through a learning lens:

• Are the learning objectives clear?
• Does the content still support the right outcomes?
• Has anything important been lost or oversimplified?
• Should parts of this become narration, practice, or performance support instead?
• Is the flow helping the learner, or just looking cleaner?

That is the real balance. Gamma can speed up the transformation, but the instructional design work is still what makes the content effective.

What Gamma does well

Gamma is especially useful when the biggest obstacle is the effort required to modernize old material. It helps reduce the tedious part of the process so the learning team can focus on higher-value decisions.

It works especially well for:

• Refreshing old PowerPoint-based training
• Turning PDFs or documents into cleaner presentation drafts
• Modernizing internal enablement and onboarding decks
• Creating a stronger first pass before deeper redesign work
• Reducing the time spent manually restyling slides one by one

What Gamma does not replace

It is worth saying clearly: Gamma does not replace instructional design, but rather supercharges it.

It does not know your audience the way you do. It does not fully understand the business context, the performance gap, or the nuance behind what should stay, what should go, and what should be taught in a different way.

What it can do is remove a lot of the friction that keeps valuable content stuck in outdated formats.

And honestly, that is already useful.

Final thoughts

A lot of learning teams are not struggling because they lack content. They are struggling because they have too much old training content trapped inside formats that are painful to update.

If you are sitting on a long training deck that still has value but clearly needs a refresh, Gamma can help you move faster. You can import the content, restructure it, modernize the presentation, and create a much better starting point without rebuilding everything from scratch.

For instructional designers, that is the real win.

Less time cleaning up slides. More time improving learning.

At SkildLabs, that is the kind of AI workflow we care about most: not AI for novelty, but AI that reduces production drag so better learning work can actually happen. You can take a look at our AI Learning category for more insightful articles.

FAQ

Can Gamma AI help instructional or learning designers refresh old training content?

Yes. Gamma can help instructional designers import older PowerPoint, PDF, and document-based materials, restructure them, and create a more modern draft to work from.

Is Gamma AI enough on its own for a learning redesign?

No. Gamma helps accelerate production, but the instructional designer still needs to shape the experience, align it to outcomes, and decide how the content should actually be taught.

What is the biggest advantage of Gamma for learning teams?

The biggest advantage is speed. It can reduce the manual effort involved in refreshing outdated decks so teams can focus more on clarity, relevance, and learner experience. This is a great introductory guide on how to start with Gamma, if your training team is considering it.

The post The Pain of Refreshing Old Training Content appeared first on SkildLabs.

That second category is where things get a lot more interesting.

In this guide, we will walk through how to create and configure a real-time AI avatar using LiveAvatar. More specifically, we will look at how to set up the avatar’s context, pull in knowledge from a website, define a persona, choose voice and language settings, test the experience, and prepare it for embedding into a site or application.

For teams working in product education, customer enablement, onboarding, or AI-first learning, this kind of setup opens the door to a much more interactive experience than a traditional talking-head avatar.

Before you dive into this particular platform, you can also check our general “How to Build Product Knowledge Training That Works in 2026” as a starter.

Why Real-Time AI Avatars Are Different

A lot of avatar tools still follow a familiar model: you pick a presenter, type in a script, generate the video, and publish it. That format is still useful, especially for explainers, updates, and repeatable content. But it is still one-way communication. The below example is from HeyGen.

Real-time avatars work differently. Instead of just presenting information, they respond to user input. That means the experience starts to feel less like watching content and more like interacting with a digital guide, coach, trainer, or support agent. This example below if from LiveAvatar.

That shift matters. If you are building learning or support experiences, the value is not only in how realistic the avatar looks. It is in whether the avatar can respond well, stay on topic, and guide the user through a useful interaction.

Step 1: Open LiveAvatar and Review the Dashboard

Once you log in, you will land on the LiveAvatar dashboard. Here you can see the available avatars, including preset options and custom avatars, depending on your account setup.

This is also where you start to see how the platform is structured. The avatar itself is only one layer. Underneath it, you are really building a combination of visual identity, context, instructions, and conversation logic.

Step 2: Understand the Role of Context

Before choosing the avatar, it helps to understand what actually makes the interaction useful. In LiveAvatar, that begins with Contexts.

The context is where you define what the avatar should know, how it should introduce itself, and how it should behave in conversation. In other words, this is where the experience becomes more than just a face and a voice.

Go to the Contexts area and create a new context.

Step 3: Add an Opening Intro and Core Instructions

Inside the dialog setup, you can define the opening introduction and the main instruction layer for the avatar. Think of this almost like the system prompt behind the experience.

This is where you can clarify things like:

• Who the avatar is
• What it is meant to help with
• What tone it should use
• What kinds of questions it should answer
• What kinds of questions it should avoid

The more clearly you define this, the better the avatar tends to perform in live interactions.

Step 4: Pull in Knowledge from a Website URL

One of the more useful parts of this workflow is the ability to add knowledge through a URL. Instead of manually copying large amounts of source material into the prompt, you can point the platform to a webpage and let it interpret the content.

Once the URL is added, LiveAvatar crawls the page and produces a summary of the information it finds. That gives you a faster way to ground the avatar in product information, service pages, documentation, knowledge hubs, or other reference material.

If you are building an avatar for product training or customer support, this step can save a surprising amount of setup time.

Step 5: Review the Generated Knowledge Carefully

Once the page has been processed, review how LiveAvatar interpreted the source. This step is important.

Just because the platform can summarize a webpage does not mean you should skip review. Look at the structure, the emphasis, and the overall accuracy of what was extracted. If needed, refine the instructions or add more context so the avatar does not drift or over-generalize.

In practice, this is often where a decent avatar becomes a much stronger one.

Step 6: Complete the Full Prompt

One of the easiest mistakes to make with conversational AI is leaving the instructions too open. A real-time avatar usually works better when it has clear boundaries.

For example, you may want to instruct it to:

• Stay focused on a specific topic or product
• Avoid answering outside its approved scope
• Offer short, direct responses before expanding
• Redirect users when the question requires a human follow-up
• Use examples only when they support clarity

These guardrails make the experience feel more intentional and more trustworthy.

Step 7: Write with Guidance

LiveAvatar also includes guided options to help you generate persona and instruction content. That can be a useful shortcut, especially when you want a strong first draft before refining the final prompt yourself.

Step 8: Save the Context and Assign It to an Avatar

Once the context is ready, save it and move into the avatar setup. At this point, you can select the avatar you want to use and connect it to the context you just created.

This is the stage where the parts start coming together:

• The avatar provides the visual layer
• The context provides the knowledge layer
• The persona and instructions shape the interaction

Step 9: Choose the Voice and Language

After the avatar is selected, choose the voice and language settings. This may sound like a small detail, but it has a big effect on the final experience.

A strong setup is not just about realism. It is about fit. The voice, tone, and persona should feel aligned with the job the avatar is there to do. A product coach, onboarding assistant, or training guide should not sound random or mismatched.

Step 10: Test the Avatar in Chat Mode

Once the setup is complete, launch the avatar and click Chat Now. The platform will start connecting the experience so you can test it live.

This is where you want to pressure-test the result. Ask real questions. Try common user prompts. Push slightly outside the intended scope. Pay attention to whether the avatar:

• Answers accurately
• Stays aligned with the persona
• Uses the right tone
• Handles boundaries well
• Feels useful rather than just impressive

That last point matters more than people think. A visually polished avatar is easy to demo. A genuinely useful avatar is what actually earns a place in a real product or learning experience.

Step 11: Use the Embed Option

Once you are happy with the setup, you can use the embed option to place the avatar somewhere else. LiveAvatar provides HTML embed code that can be copied into another site or application.

That makes it possible to bring the experience into a landing page, knowledge portal, onboarding flow, training environment, or product experience.

WordPress Embed Placeholder for an AI Avatar

If you are placing the avatar into WordPress, the easiest option is typically a Custom HTML block. Until your final code is ready, you can use a simple placeholder like this:

<!-- Replace this block with your LiveAvatar embed code -->
<div class="liveavatar-embed-placeholder">
  <p>Interactive AI avatar will appear here.</p>
</div>

What This Means for Learning and Product Teams

What makes real-time avatars interesting is not just the novelty factor. It is the interaction model.

For learning teams, this can support more immersive training, simulated roleplay, guided practice, and contextual support. For product teams, it can create a more human-feeling layer for onboarding, product guidance, and self-serve help.

That said, the value does not come from adding an avatar for the sake of it. The value comes from designing the interaction well. Good context, good instructions, good boundaries, and good testing usually matter more than the visual layer alone.

Best Practices Before You Publish an AI Avatar

• Keep the knowledge focused: Use the most relevant pages and sources, not everything you have.
• Write instructions clearly: Be specific about role, scope, and tone.
• Review generated knowledge: Do not assume the imported summary is perfect.
• Test edge cases: See how the avatar handles ambiguity, off-topic prompts, and unclear requests.
• Match voice to use case: The right voice and language settings make the avatar feel intentional.
• Design for usefulness: The goal is not just to look advanced. It is to help people get somewhere faster.

Final Thoughts

Creating a real-time AI avatar with LiveAvatar is not really about building a digital presenter. It is about designing an interactive experience.

When the persona is clear, the context is grounded, and the instructions are well-defined, the avatar becomes far more than a visual layer. It becomes a practical interface for conversation, support, and learning.

And that is where this starts to become genuinely useful.

Frequently asked questions (FAQ)

What is the difference between a scripted AI avatar and a real-time avatar?

A scripted avatar delivers prewritten content, while a real-time avatar responds dynamically to user input during a live interaction.

Can LiveAvatar use website content as a knowledge source?

Yes. In this workflow, you can add a URL and let the platform crawl the page and generate a summary that supports the avatar’s context.

What should I test before publishing an AI avatar?

Test for accuracy, tone, scope control, persona consistency, voice fit, and how the avatar handles questions that fall outside its intended purpose.

The post How to Create a Real-Time AI Avatar for Training and Onboarding appeared first on SkildLabs.