The Era of AI Creating Images, Video, and Voice: What Is Multimodal AI?

What Is Multimodal AI? The Era of Creating Images, Voice, and Video Beyond Text

Multimodal AI is artificial intelligence that can understand and generate multiple types of data at once, including text, images, voice, and video. Where AI once handled only "text → text," a single text prompt can now produce an image, generate a voiceover, and even create a video.

In this guide, B2B sales reps and office professionals get step-by-step practice with four multimodal AI tools they can put to work today.

Before You Start: Tools and Accounts You Need

Get these tools ready in advance so the exercises go smoothly.

How to Create an Account (All Tools)

1. Go to each site.

2. Click "Sign Up" or "Get Started" at the top right.

3. Sign up with a Google account or email.

4. Once you verify your email, you can start using it right away.

💡 Pro tip: Signing up with your work email makes it easier to manage the account when you share it with your team later. The free plans are enough to complete every exercise in this guide.

Exercise 1: Create Images for Proposals with AI (ChatGPT + DALL·E 3)

Work Scenario

A B2B sales rep needs a custom illustration for a proposal going to a client. Commissioning a designer takes 3 days; with AI, it takes 30 seconds.

Before / After Comparison

How to use it

Step 1: Go to ChatGPT

• In your browser, go to chat.openai.com.

• After you log in, the main chat screen appears.

Step 2: Check the model

• The model name appears at the top center of the screen.

• Make sure **"GPT-4o"** is selected. (Click to change it.)

Step 3: Enter an image-generation prompt

• Copy the prompt below and paste it into the message box at the bottom.

Prompt example 1 — Proposal cover image:

"Create a cover image for a proposal introducing a B2B SaaS solution. Make it a clean business illustration in blue tones, showing several company buildings connected to a cloud server. Don't include any text, and make it 16:9."

Step 4: Review and revise the result

• The image is generated in about 15–30 seconds.

• If you don't like it, enter a specific revision request.

• Examples: "Change the background to a deeper navy", "Replace the buildings with people collaborating"

Step 5: Download

• Click the generated image.

• Click the **download icon (↓)** at the top right.

• It's saved as a PNG file.

Prompt example 2 — Data visualization infographic:

"Create an infographic comparing before and after adopting our company's solution. The left side says 'Before: manual work, 30% errors, 8 hours processing time' and the right side says 'After: automated, 2% errors, 30 minutes processing time.' Use arrows and icons in a clean business style."

💡 Practical tip: When generating images, explicitly say **"Don't include any text"** to prevent garbled lettering. It's much cleaner to add the text you need yourself in PowerPoint.

Hands-on Step 2 — Create presentation narration with AI (ElevenLabs)

Work Scenario

You want professional narration for a video proposal going to a remote client. Hiring a voice actor costs time and money, but with an AI voice you can finish in 5 minutes.

Before / After Comparison

How to use it

Step 1: Go to ElevenLabs

• In your browser, go to elevenlabs.io.

• After you log in, the dashboard appears.

Step 2: Open the text-to-speech menu

• Click **"Text to Speech"** in the left sidebar.

• Or click the "Speech" tab in the top menu.

Step 3: Choose a voice

• Click the "Voice" dropdown at the top of the screen.

• Recommended voices: for business use, "Daniel" (male, calm tone) or "Rachel" (female, professional tone) work well.

• To find a voice that supports **Korean**, type "Korean" in the search box.

Step 4: Enter the narration text

• Paste your narration script into the large text box in the center.

Prompt example 3 — Proposal narration script:

"Hello, this is ABC Solutions. The Smart Workflow System we're introducing today transforms your business processes. Adopting companies have cut average processing time by 67%, and in one case saved roughly KRW 200 million (≈ US$149,600) in labor costs per year. Over the next three minutes, we'll walk you through the key features."

Step 5: Fine-tune the settings

• Stability slider: move it right for a steadier, more consistent tone; move it left for more emotional expression. For business use, we recommend **70–80%**.

• Clarity + Similarity Enhancement: keep the default (75%).

Step 6: Generate and download

• Click the "Generate" button at the bottom.

• The audio is generated in 10–20 seconds.

• Preview it with the play button (▶), and if you're happy with it, click the **download icon (↓)** to save it as an MP3.

💡 Pro tip: Adding commas (,) and periods (.) between sentences creates natural pacing. Putting "..." (an ellipsis) before a word you want to emphasize adds a brief pause. Example: "We saved about... KRW 200 million (≈ US$149,600) in labor costs per year."

Exercise 3 — Create a short promo video with AI (Runway Gen-3)

Work Scenario

You need a 5-second motion graphics clip to attach to a new product launch email. You don't need video editing skills — text alone is enough.

Before / After Comparison

How to use it

Step 1: Go to Runway

• In your browser, go to runwayml.com.

• After you log in, the **"Dashboard"** appears.

Step 2: Create a new project

• Click "Generate Videos" in the left menu, or **"Text/Image to Video"** on the main screen.

• Make sure **"Gen-3 Alpha Turbo"** is selected (the latest model).

Step 3: Generate video from text

• Enter a description of the video in the "Text Prompt" box.

Prompt example 4 — Product promo clip:

"A sleek modern office environment. A holographic dashboard floats above a desk, showing real-time analytics and graphs. Blue and white color scheme. Camera slowly zooms in. Professional, futuristic atmosphere. Cinematic lighting."

Note: Runway currently delivers its best results with English prompts.

Step 4: Adjust the settings

• Duration: Choose 5 or 10 seconds. (We recommend 5 seconds to save your free credits.)

• Resolution: Keep the default 720p.

Step 5: Generate

• Click the "Generate" button.

• Your video will be ready in about 1–3 minutes.

• Check the preview, then click "Download" to save it as an MP4.

Step 6: Create a Video from an Image (Advanced)

• You can also use the image you created in Exercise Step 1.

• Click the "Image to Video" tab.

• Drag and drop the image you made with DALL·E 3 to upload it.

• Describe the motion you want in the prompt: "Slow camera pan to the right, subtle particle effects floating"

• Click **"Generate"** and your still image becomes a video with natural motion.

💡 Pro tip: If writing English prompts is hard, first ask ChatGPT: *"Rewrite the following scene as an English prompt for Runway AI video generation: a modern office where data appears as holograms."* You'll get an optimized English prompt.

Exercise Step 4 — Multimodal Analysis with Google Gemini (Image → Text)

Work Scenario

You want to show AI a photo of a competitor's product brochure, a booth photo from a trade show, or a site photo from a customer, and get an instant analysis report.

Before / After Comparison

How to use it

Step 1: Open Gemini

• Go to gemini.google.com in your browser.

• Log in with your Google account.

Step 2: Upload an Image

• Click the "+" icon or the **image icon (🖼️)** on the left of the chat input box.

• Select the image you want to analyze. (A competitor brochure photo, a business card photo, a site photo, etc.)

Step 3: Enter an Analysis Prompt

• With the image attached, type your text prompt.

Prompt Example 5 — Competitor Brochure Analysis:

"This image is a competitor's product brochure. Analyze it by the following items:

1. Key product names and main features

2. Estimated target customers

3. Pricing information (if any)

4. 3 points where we can differentiate ourselves

5. Analysis of this brochure's marketing strategy

   Organize it in a table."

Prompt Example 6 — Trade Show Photo Report:

"This photo shows a booth at the IT trade show I attended today. Based on it, write a business trip report for my manager. Analyze the products on display, the technology keywords, and the market trends, and draw out the implications for our business."

Step 4: Use the Results

• Gemini analyzes the text, layout, and even the design elements in an image, then writes a report.

• Copy the result and paste it straight into an email, a report, or a Slack message.

💡 Pro tip: Take a photo of a business card, upload it, and say *“Turn the information on this card into CSV format for CRM entry.”* You can digitize dozens of cards collected at a trade show in no time.

Exercise 5 — An Integrated Workflow Combining All Four Tools

The real power comes when you connect the tools you have learned so far into a single workflow.

Scenario: Building a Proposal Presentation for a New Client

The traditional way?

• Commissioning a designer for images: 3 days

• Outsourcing the video clips: 1 week

• Voice actor recording: 3 days

• Total: 1–2 weeks and KRW 2 (≈ US$1)–5 million

Step-by-step walkthrough

Step 1: Sketch the proposal outline in ChatGPT.

• “Outline a 10-page B2B SaaS proposal. The client is a mid-sized manufacturer, and our product is a production management automation solution.”

Step 2: Generate an image for each slide concept with DALL·E 3.

• Page 1 cover, page 3 problem statement, page 5 solution overview, page 7 results case study, page 10 closing

Step 3: Capture the client’s website, upload it to Gemini, and analyze the client’s pain points.

Step 4: Turn the solution demo scene into a 5-second clip in Runway and insert it on page 5.

Step 5: Generate the full narration with ElevenLabs and insert it as per-slide audio in your PPT.

• In PowerPoint, select each slide → Insert → Audio → Audio on My PC → choose the MP3 file

Cautions When Using Multimodal AI

⚠️ What You Must Check

• Copyright: Copyright policies for AI-generated images vary by tool. Check each service’s terms of use before commercial use.

  • ChatGPT/DALL·E 3: Commercial use is allowed for paid-plan users

  • Runway: Commercial use is allowed on paid plans

  • ElevenLabs: Commercial use is allowed on paid plans

• Accuracy check: A person must always review any text, numbers, and logos in AI-generated images.

• Confidential information: Before uploading client confidential information or internal strategy documents to an external AI tool, always check your company security policy.

• Disclosure: In some cases you must disclose that content was generated by AI. Check your company’s internal AI usage guidelines.

Wrap-Up — 3 Actions to Start Tomorrow

Here is the minimum set of actions to apply what you learned today to your work tomorrow.

Action 1: Today (5 minutes)

• Open ChatGPT and create one proposal cover image.

• Just copy Prompt Example 1 above and use it as is.

Action 2: This week (30 minutes)

• Sign up for ElevenLabs and create one summary narration for a proposal you are currently working on.

• Sixty seconds is plenty.

Action 3: This month (2 hours)

• Apply the **integrated workflow (5 steps)** to a proposal for a real client meeting and complete a multimodal presentation.

Key Takeaways Checklist

• I understand the concept of multimodal AI (text↔image↔voice↔video)

• I can generate business images with ChatGPT

• I can create narration audio with ElevenLabs

• I can create short video clips with Runway

• I can analyze images with Gemini and write reports from them

• I can run an integrated workflow that combines all four tools

Multimodal AI is no longer just for technical experts. Anyone who can type text can now create images, voice, and video. Apply just one thing you practiced today to your work tomorrow. A small start can lead to a big change in how you work.