← Video Library · Workshops

Build Your Complete Video Production System: Gamma + Gemini + Descript

Published December 11, 2025 · 301 views on YouTube

Key takeaways

What does this workshop cover?

This is a replay of session five of six in Gen AI University’s Gamma Mastery workshop series, hosted by Darby. The episode walks through a live, unscripted workflow for chaining three tools into one video production pipeline: Gemini for research, scripting, and voiceover; Gamma’s Studio Mode for generating a themed slide deck; and Descript for final video assembly. The session opens with community check-ins and Q&A before moving into a live demo built around a real participant’s business.

How do you turn a service offer into a video script with Gemini?

Give Gemini a content brief: what you offer, the problem you solve, and a link to your website so it can research the business. In the demo, Darby fed Gemini a brief for a QuickBooks Desktop-to-cloud migration service and asked it to identify the problem, agitate it, and propose a solution. Gemini returned a problem-agitate-solution script (deadline crisis, liability risk, call to action to book a 15-minute strategy call) along with a scene-by-scene visual breakdown that could be handed off to a slide or video tool.

Can Gemini generate video and voiceover on its own?

Yes. A workshop participant, Mark, described using Gemini to generate full 8-second video clips with voiceover, including choosing voice characteristics like accent, age, and formality. Darby tested this live and confirmed Gemini could generate a matching voiceover clip that followed a script closely, calling Gemini 3.0 “a massive upgrade” for this kind of work. The catch: Gemini’s video generation is limited to a small number of clips per day, so it works better for short test spots than a full-length video.

How does Gamma’s Studio Mode generate a slide deck from a script?

  1. Paste the script and scene breakdown into Gamma’s Studio Mode (not Classic Mode) content brief, and let it generate an outline first.
  2. Choose the number of cards, the amount of text per slide, and a visual format (landscape 16x9 for video, or a social format).
  3. Set the art direction — either pick a standard Gamma theme or generate a custom art style guide by giving Gamma a screenshot of a brand’s existing page and asking it to identify the colors and vibe, then paste that style guide into the custom art style field.
  4. Pick an image generation model (the workshop used Nano Banana Pro) and generate. Higher-tier models consume significantly more image credits per slide — Darby noted the Ultra model uses about 120 credits per image, so an eight-slide deck can burn roughly 1,000 credits in one generation.
  5. Review and edit the output. Gamma’s “ask agent” feature can edit copy or remove elements across a slide (and sometimes across the whole deck) through follow-up prompts, but human review is still needed before anything goes out the door — the raw AI-generated slides in the demo carried stray text, off-diversity imagery, and formatting the presenter had to clean up.

How do the pieces come together in Descript?

Once slides are generated and refined in Gamma, export them as images. Record or generate a voiceover track (the workshop suggested tools like ElevenLabs as an alternative to Gemini’s native voiceover), upload the audio and the slide images into Descript, and assemble them there. Descript treats each slide as a layer with an adjustable duration (6 seconds by default) that can be synced to the voiceover, then exported for upload to YouTube or wherever the video will live. Darby described this as his own workflow rather than the only path — a participant also raised the possibility of reversing the process, starting from a NotebookLM-generated podcast voiceover and having Gamma generate matching scenes to follow the audio.

Who is this workshop for?

This session is aimed at Gen AI University community members already using Gamma who want to extend it into a fuller video production pipeline. Several participants shared their own use cases during the live Q&A, including a high-ticket sales “one-pager” decision-map framework, informational and explainer videos for client education, and marketing videos built around competitor and pain-point research from GenAI University’s Market Sauce masterclass.

Full video transcript

Hey everybody, welcome into the Gamma Mastery workshop where today we’re going to be diving into a mixture of Gamma with Gemini and Descript. Welcome in everybody. Good to see everybody this fine Tuesday. Good to see you Mike Sawson, Mark Mohamad. How’s everybody doing today?

Wonderful. How are you?

Doing great myself. Glad to hear you’re doing wonderful. Good to see you Mark. We got another Mark joining us. Let’s go. Hope everyone’s having a great Tuesday. Love to check in with everybody here who has been following along with our workshop series. How is your work going in Gamma? What breakthroughs or what progress forward have you guys been making over the past few weeks since building with Gamma? Love to hear any wins out there. Hope that we have some wins to share. If not, let me know where you guys are getting stuck.

Can I go?

Go ahead. I was wondering, I got the API and I still cannot — this is where I’m stuck — go automatically posting. That’s the purpose of getting the API, right? To link it to posting.

That is one way you can use it, but it’s not the only way.

Okay, so maybe I misunderstood or missed something.

Are you meaning posting to social media?

Yeah, yeah.

I don’t know if Gamma has a way to schedule posts directly from their app. I don’t believe they do, but you could sync up with other tools that do support the scheduling of posts.

I usually use Make.com. So maybe I’ll have to figure out how to connect them together. Thank you.

Yeah, I would check out Blotato as an example, because they have the APIs for scheduling and automating posting. Sabrina and I have been chatting about potentially doing a demo of some sort. He talks about it a lot.

Yeah. Maybe Gamma could fit into the mix there. You don’t have anything locked in yet, but that could be an alternative or an option for you if you’re using Make to set up these posts.

That’s great. Thank you.

Cool. Anybody else? Any other wins in using Gamma? Any burning questions? Is anyone stuck anywhere? Does anyone need any help with getting this stuff put into practice? Good to see you, Russell. Well, let me just ask before we dive into things today — what are you looking to get out of not just Gamma but using AI mixed with Gamma like we’re doing in these workshop demonstrations? I’d love to hear exactly what you guys are looking to create if you don’t mind sharing. It just helps me understand where everybody here is at, because next week is our last of these workshops in terms of this series for 2025, and I hate talking to a wall.

So I’d love to get some idea about where everybody here is at. Mark, I wouldn’t mind figuring out how to integrate with — sorry, not Descript — NotebookLM, to see if I can actually generate something to go with the podcast that it automatically generates.

Some sort of graphical image that rolls across my screen. I don’t know, maybe Descript is a better tool for that. I’m not sure which one is the best one, but I’ll definitely be open for guidance or suggestions on which tool to dive into first.

Cool. So is your question coming from using NotebookLM and their podcast feature, where they do an audio breakdown, but then tying in visuals to that podcast?

Yeah, that’s what I’d like to do.

Got it.

I think that’d be really cool and Gamma would probably do very well at it, especially now that it has Nano Banana.

I think so.

That’s a good idea. Thanks for sharing that.

Sorry, go ahead. I was gonna say — the other Mark had his hand raised. I’m going to go over to you, Mark. Good to see you.

So right now I’m trying to create an asset — a one-pager framework that gives high-ticket remote closers and their client companies a concrete decision map. Basically a one-pager capturing the operating criteria and the outcome criteria on the decision-making mapping that I’m doing and working out for a client right now. The problem is they’re leaving the call with “I’ll think about it” uncertainty, and it’s really because they don’t know how the thing they’re selling works — they don’t know how AI works. So essentially it’s a one-pager system we can use either on the call or after the call to give a visual and concrete one-page game plan that’s sent before or after the call to pre-frame, de-risk, create referrals — that’s what I’m creating right now.

Would this be something —

Sorry, go ahead.

Oh, a little background — I’m a high-ticket closer and I close for a couple of agencies selling AI workflows. Right now I’m trying to get back up to 20 grand a month, but I’m needing to do more on the operation side, getting the back end of what they need as an offer to make customer satisfaction numbers go up and reduce refunds. So if I set expectations up front, I feel like that might help, and get everybody streamlined as well. I created a mock one and that’s actually helping out — we’re getting fewer refunds now — and now I’m trying to create a product that can go external, that I can sell to other companies, other remote closers, other high-ticket coaching companies. Because if I ask somebody for $25,000, I kind of want a piece of it, the 10%.

Cool. So these one-pagers — would you be providing client information from a discovery call and then turning that into a one-page digestible thing? Is that the idea?

Yep. So essentially, the how-the-program-works part is the template right now, and then all the pains that you collected on the phone — what I call operating criteria. A lot of closers don’t go deep enough to know the cost of inaction, so it lands that this is what they need in order to make the thing work versus just a whole bunch of pain. Once we got through that part of the call — if they don’t think this is what they need in order to close, then your plan doesn’t really land with them. So we got over that hump. I put the operating criteria inside the one-pager, and now it’s just a matter of inserting the pain — like you said, basically funneling the Zoom call — that I already have workflows for, into a one-pager, and now I’m just trying to make it seamless for the closer.

That sounds like it could be its own product for sure. Thank you for sharing, Mark. Maybe something we can get into down the road here or outside of the workshops if it makes sense. Any other ideas, thoughts, or other ways that you guys are looking to use and leverage Gamma before we jump into some demonstration? Russell?

Yeah, I’ll jump in for a minute. Can you hear me?

Yep.

Am I good? All right. I’m just looking to create informational presentations and explainer videos for my clients and subscribers, and ultimately interactive learning courses. This last week I created a couple in Gamma and a couple of explainer videos in Google. It does an amazing job — not too excited about all the transcript it does, but I can work with that, and the amount of time it saves is amazing. I’m trying to blend those all together and create a workflow I can ultimately automate through an agent builder, through n8n or whatever. I’m swimming in the deep end. But where I want to end up is being able to automate these videos and give them a personal touch so it doesn’t sound like AI — I want to humanize it with my own touch.

Totally. Thanks for sharing, Russell. Other ways y’all are looking to leverage Gamma? Let me check the comments here. Quick question?

I guess a better way to explain it is basically creating stories with numbers. If there’s a visual that shows people where they are and where they’re looking to go, that’s probably the best thing.

Okay, cool. I think there’s going to be a never-ending list of things we can explore and experiment with in Gamma. For the rest of today’s session, I want to walk through a way to demonstrate how you can approach some of the workflows inside Gemini, maybe even jump into a little bit of Google NotebookLM — that gave me an idea on the podcasting front, mixing that in with these scripts — and demonstrate a few ways you can approach using these tools. I encourage you to experiment yourself and see how you can tie these things in. I’ve heard folks in the community looking for ways to visualize data, numbers, headlines, create stories and visual themes around content. I’d love to get an example live from people here. I can come up with ideas for whatever industry or business we have — maybe a high-ticket sales thing. I’d like to stay away from an AI selling an AI service, just for the purpose of this example — something more like a real-world service, maybe high-ticket veneer sales or an expensive home service. Does anybody work with any clients that could seed an idea for an experiment? Mark, you work with some home services. There’s three Marks on the call right now, so I just called all three of you out at once.

So I’m working on something — we just recently launched a conversion service converting QuickBooks Desktop to Xero cloud accounting. The reason people have to migrate from desktop to cloud is that QuickBooks has sunset their 2022, 2023, 2024 desktop products. They’re not selling the desktop service anymore — only cloud and an enterprise-level desktop service. So it’s forcing a lot of accountants, bookkeepers, and businesses to migrate. For individual businesses the migration isn’t necessarily that expensive, but for accountants it can be extremely expensive and a big-ticket item, because each conversion baseline starts at $299, and if they want up to six years of history or what the IRS demands, that’s another $200, and if they want 14-plus years we add another $500 on top of that. That conversion process is done across multiple files — one QuickBooks Desktop license can manage anywhere from one to a thousand QuickBooks clients. So the potential for a high-ticket sale is definitely there. That’s the product line.

Okay, and what’s your website again, Mark?

It’s quickbookswitch.com.

All right, cool. So let’s take this as an example of how we can take a service you’re providing to a specific type of people. There’s obviously money involved, there’s a chance for a high-ticket sale, and there’s a narrative — a story we can create. I’m imagining you’d want to connect with more of these accountants and help them with this migration service.

You bet.

Okay, cool. So, how I’ve been exploring using an LLM in general — context is everything here: here’s the problem, here’s what we do, here’s a little information. Gemini has been really great lately at helping with this process, and in this portion of the series we’re demonstrating starting with Gemini. You can test this out with other services too. Reaching more of these accountants — are you just wanting to book meetings with them, Mark? Is that your goal?

Yep. Bingo.

I’ve used the Market Sauce methodology, your masterclass, and I’ve got about 15 documents that go through your process, and it’s been extremely eye-opening. Our technology forced us to change a couple of the features we originally wanted to launch with, but there are still some service differentiators between us and the competitors. I’ve analyzed the competitors using your Market Sauce prompts and dived deep, and I’ve even got an initial customer-facing deck that Gamma produced for me. I definitely use Gamma weekly, sometimes daily, to facilitate the sales channels and even the design of the website and the content that went up on the splash page.

Awesome to hear that.

Then I used Descript to build a couple of videos using Gemini. I had Gemini do two eight-minute clips a day and I’ve uploaded them to the YouTube channel — it’s Wer Technologies — and they clearly articulate the pain points I developed using your Market Sauce masterclass, with actionable items. So we’ve got videos, social media posts on LinkedIn, Facebook, Instagram, all designed with Claude and the Market Sauce masterclass. Thank you so much.

You bet. I’m glad you’re putting it to work. And for those curious where you can find this masterclass, it’s actually available in our foundation training inside Gen AI University — you can head over to genaiu.com/skool and go through that entire masterclass yourself and put it into action. What I’ve done here inside Gemini is give it some initial context around how I’d go about solving this if I were a service provider looking to create some of these assets. I started by asking Gemini for a basic content brief — this is what we offer, here’s some of what’s happening right now, here’s a link to my website, I want you to research my site and the problem I’m solving. So Gemini identified the problem, agitated it, then provided a solution back to quickbookswitch.com. It says: the QuickBooks Desktop crisis is here — if your firm still manages a portfolio of QuickBooks Desktop clients, you’re facing a massive liability, an unmovable deadline, dozens or hundreds of client files to migrate, and you can’t afford to get this wrong. Then it moves into agitating the problem a little more. Mark, being the expert in this case, can you read this?

This is accurate. 100% accurate.

Yeah, calling that out because when anybody’s working with AI, it really helps to understand the field you’re working in, because AI can hallucinate, and you want to make sure it’s on the right track. So with this problem-agitate-solution, the call to action is: protect your reputation, reclaim your firm’s capacity, secure your client data, book a 15-minute conversion strategy call with us. From this initial script, we have the scene-by-scene visual story for a video — deadline crisis, scale up the volume, option for the risk — it even gives ideas for overlays and things like that. Then a call to action, clear switch, voice over: protect your reputation, reclaim your firm’s capacity, book a 15-minute call. So I’ve got the starting points — if I wanted to make a quick video spot or commercial, I could follow this voice-over script and either record it myself or run it through a tool like ElevenLabs. Mark, have you been doing the voiceovers yourself when you’ve gone through this process?

I’ve allowed Gemini to do it and it does a great job, especially with ethnically diverse voice actors.

This is new to me — you’re telling me Gemini can actually generate voiceovers?

Yeah, really.

We’re learning everything right now in real time. I did not realize Gemini could do voiceovers. Did anybody else know that? That’s pretty cool.

If that’s the case, check out the two videos on the Wer Technologies YouTube channel — they’re not embedded here, you’d have to go to YouTube.

Let’s stay inside Gemini right now and go through this process a bit. I’m assuming the prompt is just “can you create a voiceover following this exactly”?

Did you give it a preference for a male or female voiceover?

I gave it — for example, the videos I gave two situations. One was like an Alcoholics Anonymous-style meeting except for accountants reviewing all the pain points of what this transition meant to them, and that’s formal — most people in those videos are dressed in suits and ties. I gave it specific nationalities, ethnic qualities, age groups, and tried to make it as diverse as possible, and it did quite well. The other one is accountants meeting at a coffee shop discussing this problematic issue — four scenes strung together with different accountants each scene, all having coffee and muffins. Gemini does an amazing job with these short videos, but you only get two a day. That’s the problem.

Was it generating the video with that, or was it an audio file you were creating?

Video with audio, the whole thing, and it does a fantastic job.

Okay. Let me see. I’m so impressed with Gemini’s new features — Gemini 3.0 is a massive upgrade.

It’s pretty impressive. I’ve known you can do audio with video — I was curious if you could just generate the audio file from these, because you can tell it to generate a video and it’s going to create one video. I think if it can’t do a full script on the audio front, options people can use are tools like ElevenLabs, or your own voice over. You can also hire people for this. ElevenLabs is magnificent at creating these voiceover tasks, and I think a lot of people doing voiceovers are going to find themselves without work shortly.

Yeah, that’s a pretty big conversation to have — being a voice actor right now seems like a dangerous field to have all your eggs in. Not that I’m advocating for voice actors not to make money, but there’s a lot of adapting going on in that part of the market. Right now, in this example — create a video for this — I asked it to research the most optimal voice that would resonate. So it gave voiceover recommendations: mid-career, professional, authoritative, clear, reassuring. Let’s see how well it follows the script, because these generate in about 8-second clips right now.

Stop wasting time on complex migrations and high error rates.

Switch to QuickBookSwitch for 99% accuracy. Book your 15-minute call today.

Not a bad little 8-second spot. Could you guys hear that?

We heard it when you played it.

I heard it. Okay, cool. So Gemini, for those who didn’t know, you can generate videos inside it, and now videos can include voice over, which is pretty cool. There are going to be some limitations on how many videos you can edit or create. We don’t have all the time in the world today to generate every individual type of video, but as an example, following this script and framework and flow, we can use it inside Gamma to generate a presentation, and then start building it out inside Descript scene by scene. We’ve got the full script outlined right here.

Once you’re in something like Gemini, you do research, you come up with a basic script for a video — say the goal is to have it book a call — you have the context of the video and a scene-by-scene visual for that video. I’m going to copy all this info and open up my Gamma. I’ve been experimenting a lot more with Studio Mode for generating different presentations and scripts. Depending on how many scenes you’re wanting to use, you can generate a certain number of cards, add different levels of text onto them. Classic mode versus Studio Mode is the difference here — Classic goes to the prompt editor, Studio Mode is different. I’m running into some issues with Gamma not going right into Studio Mode to generate the content, so let me pause and see what’s going on. Has anybody been using Studio Mode for generating slides? I want to get a good example of that generation because it’s significantly different from Classic mode.

Okay, I think there’s a technical thing happening right now, because when you’re in Studio Mode, this is how it usually looks — you describe something you want to make, and it generates an outline based on what you put in. The issue I’m running into is when I paste the text, it’s going into the edit prompt instead of generating the outline. Let me try regenerating the outline — maybe that found the issue. Sometimes the tech just doesn’t work the same way every day.

So when you’re in Studio Mode, you can select the number of cards, select Studio as the default, and you can do this for social-media-style content as well — I’ve usually been using it for landscape, traditional 16x9. It generates an outline based on the information you provide in the context brief, and that outline is what gets generated inside the studio. Before you generate content in Gamma Studio, you can also select the amount of text you want on any slide, and select a theme or visual. Let me quickly look at Mark’s page — it’s green and blues. So I’m going to give it a screenshot of the page and say: create a custom art style guide that matches the theme of this page. The reason I’m asking for a custom art style guide is this is where you’ll probably spend more time if you’re creating more of these Studio-style decks.

That’s not what I was asking for — I don’t want a full image in Nano Banana, I just want the prompt. Okay, this is what I was wanting to ask for — you give it a page, a landing page, whatever you want to analyze, it identifies the colors, the themes, the vibe, and you get an art style guide like this. You can come back and test different versions to see how it plays out. When you copy this, you add it to the bottom section where it says art image / art style guide, select custom, and paste it in. You can add additional keywords too — if you want a specific vibe, like a glass aesthetic, which is pretty popular these days.

If I were Mark and had a Gamma account, I’d spend time creating a custom theme for my business, but you can also start with standard Gamma themes. Let’s go a green-and-blue route — Pistachio is maybe a little closer. I’ll make it a little techy, which gives it a general direction. If you have your brand colors set up, you’d want to review the amount of text you want on the slides — options range from minimal to lots of text. Nano Banana is really good at adding words nowadays and keeping the text spelled correctly, though it still goofs up sometimes.

Let’s say I give this outline a thumbs up — QuickBooks Crisis, Option A, The Failure Rate Gamble, and so on — this is the outline, theme, and art style guide I want to follow. Be careful with “auto select” for the image model, because if you’re still experimenting you may want to try multiple models. Be aware of the image credits this uses — on the Pro plan you get around 4,000 credits a month, and the Ultra model is about 120 credits per image. With eight slides, that’s roughly a thousand credits on one generation — you can burn through credits very quickly if you’re not paying attention. That happened to me the other day. For this example I’m going to select Nano Banana Pro rather than the newer HD model, since I haven’t played with HD much yet, and expect around 80 credits per image. I’ll click generate.

Now we’re using Gamma Studio Mode, which is much more image-heavy than Classic Gamma, which leans more on editable text with images incorporated into slides. Studio Mode generates images for the slides, which takes longer and costs more credits. I encourage everyone using this to spend time on the creative direction up front, so you have the best shot at producing quality slides. If you switch tabs while it’s generating, it often just stops and you have to regenerate. It might take a minute or two. As it comes up, we can see QuickBooks Desktop Crisis, your firm’s allowed 3,500 liability — I’m pretty impressed with how good Gamma can produce all these images in one batch and one consistent style.

This can get modified into a lot of different art styles, and you can see where some of this is a little off, maybe because I gave it too much context — there’s no reason to put some of this in a commercial for your business, Mark, no one needs to see this. We can delete that page. But this slide may be a good one to work with.

The only problem is sometimes Nano Banana produces an ethnic crowd that’s not quite mixed — it seems to have done okay here, but I find I often have to go back and tell it to show age and ethnic diversity in these clips. This one seems a bit more diverse than a crowd filled with just one type of person, but that’s something you can guide with prompting when you’re using a tool like Gamma that taps into these image models. It’s a good thing to be aware of on any front, since AI can skew one way or the other in who it shows.

So here’s our first generation of slides and images, given we’re limited on time today. This is where human-in-the-loop review is a necessity — if you just published this as-is, it would look silly to tell people this is your offering with the art style guide text still on it, so cleaning it up and even removing certain slides or pieces of a slide is necessary. One thing you can do inside Gamma is go to “ask agent” and ask it to remove certain copy — remove the headline about the art style guide, for example. It may take a couple of prompts back and forth, and you can edit images directly inside Gamma to refine content and copy. It removed the headline as asked, though it may leave the “communicating trust” text since that wasn’t specifically called out, and it left a bit of blank space.

Is it possible to ask it to edit the whole document and remove those references, or do you have to go slide by slide?

We can ask it — there’s probably some limitation to what the agent can do; it usually focuses on the selected slide, but it looks like every one of these images got edited at once, which is impressive. That’s one of the things I’ve been impressed with in the new Gemini — it seems to edit itself quite well, not on complex diagrams, but definitely on images and overall content. Until Gemini 3 I didn’t use it much, but it’s a game changer, especially for creatives.

This made a pretty significant amount of editing — you can look at it side by side, the original and the modified, and it’s a lot cleaner. It also duplicated a few things and didn’t perfectly incorporate the body font style and accent colors, but it kept the overall images, style, and flow following the custom theme and colors we set up in Gamma.

From here, we could spend more time refining and reviewing, even taking individual slides back into Gemini directly to get really specific about how a slide should look. But let’s assume we spend another 30 minutes or a few hours getting these completely perfect. We can take this and turn it into a quick 15-to-30-second video by exporting the slides — I’d export these as PNGs if I were building a video with a voice over added afterward. We’ll imagine the voiceover script: “The QuickBooks Desktop crisis is here. If you’re a QB Desktop client, you’re facing a massive liability and an unmovable deadline.” You write out the script like that.

The next step would be taking that script, recording the audio first (say, through ElevenLabs), uploading that audio into a Descript file along with the image slides, and creating the actual video inside Descript. When you upload slides in order in Descript, you can add different layers to them — right now the only layer is the slides themselves, but you can adjust how long each slide displays; by default they’re 6 seconds each. You organize the slides to match the voice over, then export and upload directly to YouTube or wherever the video will be seen.

There are obviously a lot of moving parts to tying these tools together, but to sum up — we started with Gemini, moved into Gamma to build a series of slides, and then finalize the video by putting the finishing pieces together inside Descript, at least in my workflow. That same approach can be used to visualize data in different ways — you can use Classic Gamma for how you want data to show. For me, Descript is the final piece that ties everything together once you’ve already created the audio, video, B-roll, voice over, and music — it gives you a workspace to bring it all together, though there’s a lot of prep work that goes into everything before it reaches Descript. I know Descript now integrates with tools like Nano Banana and others, but I haven’t explored that deeply yet. For me it’s a pretty straightforward flow: whatever LLM you’re using to build out the research and outline, moving it into Gamma, and finalizing the video in Descript. That’s by no means the only workflow — you could also go through NotebookLM, extract podcast voices, upload those into Descript, and layer video and images that the voices narrate.

Is there a way to reverse that process — take a NotebookLM podcast and put it into some sort of slide generator, having the voice be the lead, and then generate visuals based on that script and upload it into Gamma? Is that a possibility?

Taking a voice over that NotebookLM generates and using it to create scene-by-scenes to follow and match the voiceover?

Yeah, then load that up into Gamma and have Gamma generate those scenes.

Yeah, I took a similar approach — I didn’t use NotebookLM, but mapped out a script for a client that the client spoke, and once the script was finalized we created scene-by-scene visuals to match the cadence of what he was saying. So it seems like a viable way to go — reverse-engineering it, “here’s the full script, create a scene by scene that matches it,” then inside Gamma get particular about which scene you want. Most of my time on a project like that goes into syncing up the beats and how the voice overlays with the different cut scenes. You could go the other way around with it — ultimately it’s about whether the visual matches what the story is telling, has a good cadence, and makes sense to people following along.

I’ve noticed a lot of videos popping up on my YouTube feed that are AI voice with AI images, and I’m not sure exactly what steps or tools they used, but ultimately that must be possible using the tools described here today.

Yeah, there are a lot of other tools you could use to approach this, and some may tie these steps together so you don’t need to move outside of it to something like Descript. This is just a workflow I’ve been using myself for the projects I’m working on, and it works for me. I hope everyone watching gets a sense of how you could work inside Gemini, starting with research — it all stems from what context you bring in and how you want to direct it. Mark, you’ve clearly done a lot of research and thinking around how you want scenes laid out, what kind of people should show up in them — components a lot of people won’t even consider, taking whatever AI gives them the first time even if it doesn’t hit the mark, because the creative direction wasn’t thought through to the degree of someone who spends extra hours mapping it out, going through the Market Sauce masterclass, understanding the target audience and what resonates with them. There’s a lot that goes into this before you ever touch Gamma. To me, that’s how you get better content out of a tool like Gamma — by doing the work that feeds the context into it.

Hope that example was helpful to watch through. We’ve got a few minutes left — any questions, thoughts, comments?

No, I’m good — you covered things really well, and I like Mark’s suggestion. A question for Mark — what’s your YouTube channel? You said you have videos showing the process.

It’s Wer Technologies. I’ll get the URL and share it in the chat here.

Thank you.

I’m going to do a quick sidebar for the other Mark — I think that was Baldwin, about the one-pager framework.

Yep, that’s it.

So, something to think about that might help you build that process out inside Gamma — have you tried playing around with their templates yet, creating your own template?

I did, yeah.

Was that useful in getting the structure you’re looking for?

Definitely useful for the structure. Correct.

Sorry, it was —

Yes, it was definitely useful.

Okay, cool. That’s all I wanted to say — having a template set up would probably help with some of that process, but for productizing it, there’s probably something at the next level with API integration that could help.

Yeah, thank you, I appreciate that.

Cool. Anything else while we have some time left today? This was number five of six in the Gamma Mastery series — we started pairing it with Gemini and other tools this month in December, but next week is our last week of doing this for 2025. Not sure exactly what we’ll cover next week, but ultimately you’ll probably learn a lot more by going and creating stuff outside of these workshops, playing around with Gamma, putting it to the test, and finding out where you run into limitations or where it’s crushing it for what you’re looking to create. Regardless of where you start, it’ll help to have your own theme set up in Gamma so you’re creating content around the same color, brand, and style you want, then using that theme, building out your own templates, and configuring the API for more automation if you’re at that level — and sharing where you’re at inside the community, so we can support each other in getting the most out of these tools. Anything before we wrap up?

That’s great, thank you, Darby.

Thank you. Much appreciated, Darby. Appreciate this work.

Getting better at it.

Glad to hear that. All right, y’all, we’ll see you inside the community. Have a great rest of your week.

All right.

Watch on YouTube

Related tutorials