The Ultimate Guide to Creating AI Influencers and Viral Content · Both Methods in One Engine.

In the club, we hosted two sessions, each of which alone is worth gold. Adi Bergman's session taught how to think and write viral content: how to stop the scroll in three seconds, how to write from the end, and how to build a complete visual world from a single master shot. The AI influencers session taught how to produce: how to create a realistic character in Nano Banana Pro, place a product in her hand, animate in Google Flow, make her speak Hebrew, and connect scenes into a long video. This guide unifies both into a single engine. One side is the head: idea, script, emotion. The other side is the hands: images, video, speech, editing. Most people do only half. Here you get the entire pipeline, from the idea to the network-ready video.

If you have already read the two separate guides, this guide is the unified workflow that connects them. If not, this is a great place to start: everything you need is right here, ordered in the actual sequence of work.

What you will be able to build by the end of the guide

• An idea that passes the scroll test using the Moday method.
• A script written from emotion and from the end, not from a blank document.
• A realistic AI influencer character with consistent identity.
• A master shot that defines a world, and from it angles, close-ups, and emotional states.
• Integration of a real or invented product in the character's hand.
• A character speaking Hebrew and presenting the product.
• A video longer than 8 seconds using clip stitching and frame continuity.
• An edited and final output from CapCut, ready for Reels, TikTok, or Stories.

The core insight: two halves of one machine

The unified pipeline · An overview of the entire process

• Head stage: Choose a human situation with real friction and test it using the Modae method.
• Writing stage: Build a matryoshka of documents and reach a short, dense script.
• World stage: Create a single master shot that defines the scene.
• Character stage: Create a realistic character in Nano Banana Pro with a consistent identity.
• Breakdown stage: Break the world down into angles, close ups, and emotional states.
• Product stage: Integrate a product into the character hand.
• Video stage: Animate in Google Flow / Veo 3.1 or in Kling using a first and last frame.
• Speech stage: Make the character speak in Hebrew, or add lipsync in HeyGen.
• Continuity stage: Connect segments into a video longer than 8 seconds.
• Editing stage: Finish in CapCut and export to social media.

Part A · The Head: An Idea That Stops the Scroll

4 conditions for a strong opening

• A familiar concept in the first three seconds: product, holiday, work situation, everyday moment.
• A clear emotion: laughter, excitement, annoyance, nostalgia, embarrassment, or relatability.
• A promise that there is more to wait for: a first punchline that hints more punchlines are on the way.
• A reason for a second view: something funny, dense, or surprising enough to return to the beginning.

Prompt · Idea testing using the Modea method

I want to test if an idea for a short video can stop the scroll. The idea: """ {Paste your idea here} """ Analyze it based on 5 points: 1. What familiar concept to the viewer pops up in the first three seconds? 2. What emotion does the video try to evoke? 3. Does the viewer have real friction with this concept in life: money, time, energy, relationships, food, work, children, status? 4. What is the first punchline or surprise that can hold them for 3 more seconds? 5. What is missing so people would want to watch it a second time? Do not write a script yet. Only a sharp and direct diagnosis.

Part A · Writing: A Matryoshka of Documents

The matryoshka layers

• Personal document: free pouring of everything you think and feel about the topic, without censorship
• Ideas document: extracting real directions for the video out of the mess
• One-sentence definition: a single one-liner that defines the video and serves as a compass
• Punchlines: jokes, moments, metaphors, and exaggerations, before the script
• Logline: what happens after what, the sequence of events
• Script: only now turning everything into dialogue or voiceover
• Production appendix: what needs to be created, animated, recorded, and edited

Prompt · Building a matryoshka of documents

I am working on a short video for social media. The initial idea: """ {Paste your idea here} """ Help me build a matryoshka of documents: 1. Ask me 7 questions to help me write a personal document on the topic. 2. After I answer, extract 10 possible ideas for the video. 3. For each idea, write one clear one-liner. 4. For the strongest idea, suggest 20 punchlines or comedic moments. 5. Then build a logline: a sequence of what happens after what. Important: Do not try to be funny by force. Do not write cheap wordplay. Look for human situations, embarrassment, exaggeration, conflict, and real friction.

Prompt · Expanding a concept world into raw materials

I am writing a video about the following world: """ {For example: skincare product / donuts / gym / mothers / holiday} """ Give me 100 concepts, situations, objects, phrases, habits, small pains, and conflicts that people know from this world. Divide the list into categories: 1. Physical objects 2. Phrases people say 3. Awkward moments 4. Small pains 5. Things that can be comically exaggerated 6. Visual imagery Do not write jokes. Provide raw materials for writing.

Adi's principle: AI is great at raw materials, less good at the final punchline. Ask it for 100 concepts from your content world, and then you choose what sparks an idea. Human taste remains yours.

Part B · The World: Master Shot Before Everything

Prompt · Master shot in Nano Banana Pro

Create a cinematic master shot for the following short video scene: Scene: {Describe the scene here in simple sentences} Requirements: - Vertical 9:16 composition - Realistic but slightly heightened cinematic look - Clear readable characters and body language - Natural lighting that matches the scene - No text, no captions, no logos, no watermark - The image should feel like the main establishing shot of the scene - Keep the environment coherent and practical for later close-ups Style reference: {If you have a reference from Pinterest or an existing image, upload it alongside the prompt}

Prompt · Breaking down a master shot into 15 angles

This is a master shot image of a scene I want to turn into a video. Create 15 prompts for me to generate additional camera angles from the same scene: 1. Five special cinematic angles that express emotion or tension. 2. Five angles focusing on the main character in different sizes: long shot, medium shot, close-up. 3. Five angles focusing on the second character or an important object in the scene. In each prompt: - Maintain the same world, same lighting, same clothing, and same perspective relative to the master shot. - Do not change the visual style. - Do not add text or subtitles. - Write in English, short and clear, suitable for Nano Banana Pro.

Part B · The Character: Realistic Influencer in Nano Banana Pro

Prompt · Realistic UGC influencer

Create a photorealistic vertical 9:16 UGC-style selfie photo of a young Israeli woman in her early 20s, casually filming herself on a phone in a real apartment kitchen. She has natural skin texture, realistic facial features, slightly imperfect lighting, and a casual everyday outfit. The background should feel lived-in and local, not like a studio: small kitchen details, soft daylight, slight visual imperfections, realistic phone-camera look. She is holding an iced coffee and looking directly at the camera as if she is about to recommend something to her followers. No text, no captions, no logos, no overly polished fashion-shoot look.

The secret to realism: do not go for the prettiest. Too perfect of an image looks like AI. Imperfect lighting, a slightly messy background, a local look, and photography that feels like a phone all make it pass better to the eye.

Part B · The Product and Emotions

Prompt · Integrating a product into the character's hand

Using the influencer image as the main reference and the product image as the product reference, create a new vertical 9:16 image where the same woman naturally holds the product in her hand, close to the camera, as if she is about to talk about it in a casual UGC video. Keep her identity, outfit, lighting and kitchen background consistent. Make the product proportions realistic. Make the hand grip natural. Preserve the exact product packaging as much as possible. No text overlays, no captions, no extra logos.

Prompt · Set of emotional states for the character

Using the attached character image as the exact identity reference, create the same character in the following emotional states: 1. Neutral 2. Angry 3. Worried 4. Sad 5. Excited 6. Confused 7. Suspicious 8. Embarrassed 9. Deeply moved Keep the same character identity, same clothing, same visual style, same lighting and same background. Only change facial expression and subtle body language. No text, no captions, no logos.

Pay attention to proportions. If you upload a very long product image, the model sometimes changes the aspect ratio of the entire image. Always add vertical 9:16 to the prompt, and if the image comes out distorted, ask for a correction immediately.

Part C · Video: First Frame and Last Frame

Prompt · Transition between first frame and last frame

Use the first image as the starting frame and the second image as the ending frame. Create a smooth, realistic video transition where: - The motion between the two frames feels natural and continuous - Human hand and body movement look real, not animated - The revealed object or product appears gradually and believably - Keep the camera angle, lighting, environment and identity consistent with the two reference frames - The motion should feel like real footage - No text, no captions, no extra objects Keep it vertical 9:16.

Part C · Speaking Hebrew and Continuity in Google Flow

Prompt · Flow: Influencer speaking Hebrew about a product

The woman is filming a casual selfie-style UGC video in Hebrew. She speaks naturally and emotionally, like she is telling her followers about a new product trend she just discovered. She starts with the first frame and smoothly transitions into holding the product shown in the final frame. Her facial expressions should be excited and believable, with natural hand movement and casual body language. She occasionally looks at the product and then back at the camera. Keep the same identity, same outfit, same kitchen background and same vertical 9:16 phone-video style. Make the Hebrew speech sound as natural as possible.

Prompt · Long video scene continuation

Continue the same selfie video from the previous final frame. The woman keeps speaking in Hebrew with the same voice, same personality and same excited UGC tone. She now puts the product aside, picks up her phone, and reacts as if she is showing something she found online. Keep the movement smooth and natural. Keep the same kitchen, same outfit, same lighting and same vertical 9:16 style. Avoid sudden camera jumps or changes in identity.

Part C · Lipsync for complete speech control

Prompt · HeyGen Motion

Animate this character naturally while speaking. The mouth should move fluently and accurately with the audio. Add subtle human facial expression, eye movement and small head motion. Keep the background stable, but allow subtle natural movement in background elements so the shot feels alive. Do not exaggerate the motion.

A small detail that makes a big difference: when there is subtle movement in the background, people walking by, changing light or a living backdrop, the video feels more believable. A frame where only the mouth moves and everything else is frozen feels artificial.

Part D · Editing and stitching in CapCut

The full pipeline · End-to-end execution list

• Choose a human situation with real friction
• Test the idea using the Modia method
• Build a nesting doll of documents down to a short and dense script
• Create a master shot that defines the world
• Create a realistic character in Nano Banana Pro with a consistent identity
• Break down the world into angles, close-ups and emotional states
• Integrate a product in the character hand
• Prepare a first frame and last frame for each segment
• Animate in Google Flow / Veo 3.1 or Kling
• Add Hebrew speech in Flow, or lipsync in HeyGen with recorded audio
• Connect segments using frame continuity into a long video
• Finish in CapCut: pacing, cuts, punchlines, music and export

AI Master club opens the door for you: live meeting every week, recordings and community. ₪47 per month, cancel anytime.

When to use each track

The complete toolkit · All tools from both sessions

ChatGPT

Thinking, idea development and expanding the conceptual world

An excellent tool for the early stages: testing an idea according to the Modia method, lists of concepts from the content world, and phrasing prompts for other tools. Ask it for raw materials, not a ready-made viral video.

Tip for creators: ask it for 100 concepts, conflicts and situations from your world, and then choose what sparks an idea.

Claude

Developing depth, structure, script and strategic thinking

Suitable for breaking down an idea, building a nesting doll, writing a logline and improving a script. When you need depth and broad context, this is a very strong tool.

Tip for creators: give it the one-liner, punchlines and logline, and ask to improve structure and pacing, not to invent from scratch.

Gemini / Nano Banana Pro

Creating characters, master shots, product and emotional states

The image model inside Gemini. The heart of visual production: realistic character, master shot, product integration, set of emotions and video frames. You can work with it using a dedicated Gem that receives a reference and returns a prompt.

Tip for creators: first create one strong image that defines a world or character, and only then break it down into angles, product and emotions.

Google Flow

Animation, Hebrew speech and connecting frames with Veo 3.1

The central tool in the production track. Creates images, turns first and last frames into video, animates Hebrew speech and connects scenes. Still experimental, but very strong.

Tip for creators: ensure Portrait 9:16 before any creation, and start with one result to save credits.

Veo 3.1

Google video model inside Flow

The model that animates the character and scene inside Flow. Good for testing against Kling and Seedance, because the same input can behave differently in each model.

Tip for creators: give it strong frames. The higher the quality of the end images, the better the video comes out.

Kling

Image animation from first frame and last frame

A strong video tool for motion between frames. Enters at the stage where you already have clear visual anchors and want the model to complete the motion. Sometimes yields better motion than Veo.

Tip for creators: do not settle for a text prompt. Provide strong first and last frames.

HeyGen

Lipsync and animating a speaking character from an image

The track for precise speech, especially in Hebrew. Upload an image and audio, and the tool creates mouth and facial movements. Also the best way to work with your own real character.

Tip for agents: In Hebrew, record the sentences in your own voice with the right emotion, and then use that for lip-syncing.

ElevenLabs

Voice generation, voice cloning, and dubbing

An audio tool for voice generation, dubbing, and narration. Suitable after you have recorded an intonation and want to replace it with another voice while preserving the pace and emotion.

Tip for agents: Keep the human intonation. The final voice can change, but the pace and emotion matter more.

MiniMax

Voice, video, and additional creative models

An AI platform with models for voice, speech, and video. An alternative for voice cloning and voice generation, especially when you want more flexibility.

Tip for agents: Test it against ElevenLabs on the same audio file and choose based on naturalness in Hebrew.

Seedance

ByteDance video model for scenes and commercials

A video model for generating videos from a description, image, or references. Relevant when you want to turn a storyboard or product image into a commercial or short video.

Tip for agents: Use it after you have a clear storyboard. The more organized the input, the less random the video.

Google AI Studio

Advanced work with Gemini and 4K images

Google environment for more technical work with the models. Also the place to get 4K quality images and connect image generation to an automated system.

Tip for agents: Move here when you want to connect the process to an automated system instead of working manually every time.

KIE.ai

A single interface to run image, video, voice, and LLM models

A convenient and relatively cheap way to access various models from one place, instead of opening an account in each tool separately. Especially suitable for building automations.

Tip for agents: Useful when building an automation with Claude Code or Codex and you want a single API that centralizes several models.

fal.ai

Access to video models via API

A service for running various models via API, including ByteDance video models. Suitable for anyone who wants to connect the process to an automation or a system that builds videos at scale.

Tip for agents: In an automated system, fal.ai can be the layer that connects images, prompts, and video.

CapCut

Editing, cuts, sound, subtitles, and finishing

An accessible editing tool for mobile and desktop. After all segments are ready, this is where you connect, cut, add subtitles and music, and export to networks.

Tip for agents: Even if everything was created in AI, the editing is where the video becomes funny, fast, and precise.

Claude Code / Codex

Automation of the entire pipeline

Code agents capable of building a system that receives an image or idea, creates prompts, activates model APIs, and returns an asset package. The direction of upcoming meetings: turning the manual workflow into a single button.

Tip for agents: Suitable when you already have a working manual workflow and you want to turn it into an automated system.

The part that cannot be fully automated

Common mistakes to avoid

• Opening an AI tool before having a human scroll stopping idea
• Writing a script from a blank document instead of building a matryoshka
• Creating a character that is too pretty and perfect, which looks less credible
• Skipping the master shot and generating each image separately, until the world is inconsistent
• Settling for a single image for a character instead of a set of emotional states
• Asking to create a video instead of providing a first and last frame
• Giving the model Hebrew text that is too long instead of short sentences
• Trying to create a long video as a single segment instead of connecting segments
• Skipping editing in CapCut and relying solely on the raw output

Ideas for business use cases

• UGC video for a physical product with an AI influencer speaking Hebrew
• A series of comedic narrative videos around the brand
• An AI presenter introducing a new service or feature
• Short training videos for employees
• A/B testing of several characters, angles, and messages
• Product commercials for clients without a shoot day and without actors
• Regular network content without being a bottleneck

Frequently asked questions

No. For a narrative video, the entire writing and master shot path is worth it. For a product UGC video, you can shorten the writing and invest in production. Choose according to the content type.

Flow and Veo 3.1 animate scenes and also understand spoken Hebrew. Kling is strong in motion between frames. HeyGen specializes in accurate lip-syncing from an image and audio, and also fits your real persona. They are often combined.

You can shorten the text and avoid words the model stumbles on, or work via the lip-sync path: clean Hebrew audio in ElevenLabs or via your own recording, then lip-syncing in HeyGen. As time goes on, Hebrew in the models is improving.

Two things. At the content level: a human idea that stops the scroll. At the production level: image quality. A precise first image, product image, and final frame lead to better video.

A second view. When someone comes back to watch the video again, it is a sign that you pulled them out of mindless scrolling and made them choose your content.

Large parts, yes, with Claude Code or Codex and an API like KIE.ai or fal.ai. But the taste, comedic choices, and final editing remain with the creator.

This is the complete pipeline from both sessions, in one place. In the AI Master club we continue from here into implementation: prompts, automations, exercises, and tools you can bring into your business or content this week.