Edit & Personalize with AI Prompts
Learn how to use Supademo MCP to edit and personalize demos at scale with natural language prompts.
About this lesson
Learn how to edit and personalize product demos at scale using natural language prompts through Supademo MCP (Model Context Protocol). Instead of making repetitive manual changes one step at a time, you'll discover how to connect your AI assistant to your Supademo workspace and use targeted prompts for bulk design changes, personalized copy and voiceovers, and automated tracking link generation.
Key takeaways
- What MCP is and how it gives AI models secure access to your Supademo workspace
- How to set up Supademo MCP in Claude or other compatible LLMs
- Three practical prompts: bulk design changes, personalized text & voiceovers, and tracking link generation
- When to use MCP vs. the built-in visual editor for different types of edits
Prefer to read? Expand to read the lesson
Intro and agenda
In this video, we give you a tactical overview of how to edit and personalize product demos with natural language prompting, using Supademo MCP.
We cover four things: what MCP actually is and why you'd use it instead of doing everything in the Supademo UI; how to set it up in your LLM like Claude; three practical prompts you can use right away; and how to think about when to use the MCP versus when you should still jump into the built-in visual editor.
The three prompt examples cover:
- Making bulk design changes
- Rewriting and personalizing text and voiceovers
- Generating personalized tracking links from a spreadsheet of leads
What is Supademo MCP?
At a practical level, the MCP, or Model Context Protocol, gives compatible AI models and agents a secure, standardized way to access and work with demos on your Supademo workspace.
That means you can access, edit, and act on demos in your workspace through Claude or ChatGPT, using natural language instead of manual step-by-step editing.
For example, maybe you want to apply zooms across dozens of steps. Maybe you want to restyle every hotspot in a demo. Maybe you want to rewrite every step for a new audience using AI. All of this can be done in bulk in seconds, using the MCP. And for fine-grained, specific visual changes? You still have access to the built-in Supademo editor for precise control.
Prefer staying in Supademo? AI Command runs the same actions from a panel inside the dashboard — no setup required.
Tips
- Be specific — include demo IDs, exact color values, and scope boundaries.
- Ask the AI to summarize changes in a table before applying them.
- Share reference links, docs, and branding guides as context.
- Test prompts on a non-critical demo first.
Common Pitfalls
- Don't rewrite live demos without reviewing proposed output first.
- Use the built-in editor for fine-grained, pixel-perfect changes — not MCP.
- Always define explicit scope to prevent unintended edits.
Recommendations
- Use MCP for bulk styling, copy rewrites, and mass link generation.
- Use the editor for final QA after MCP changes.
- Prompt the AI to propose changes in a table before applying.
- Leverage dynamic variables to personalize one demo for many recipients.
Setting up MCP in Claude
Here's how to set up the MCP in Claude on the web:
- Open Claude and log into your account
- Click your profile icon in the bottom-left
- Go to Settings, then Connectors
- Click Add custom connector and enter the Supademo MCP URL and name
- Authenticate with your Supademo workspace
- Open a new chat and start prompting
One tip before you jump in: the more specific, contextual, and well-planned your prompt is, the better your result will be. Be specific about what demo you want to work on, what you want changed, and even ask it to summarize the changes to review before it actually makes edits. Feel free to give it reference links, docs, branding guides, or other material to take in as context as well.
Use case 1: Bulk design changes
Instead of changing zooms, hotspot styles, colors, and branching steps one step at a time, we're going to ask Claude to do it in one pass. Here's the prompt:
Example prompt:
I want you to update this Supademo: [INSERT DEMO ID OR URL] with the following changes: 1. Apply zooms across every step, using variations between 75% and 100% depending on the focal area around the hotspot. 2. Update some of the hotspots to use backdrop-dark to create some variations. 3. Change hotspot colors to: - primary hotspot color: #65a30d - label background color: #eab308 4. Update the chapter background to this attached image and change the primary button color to #65a30d. 5. Add a secondary button labeled 'Book demo' that navigates viewers to the last step, with bg color #eab308 and text color #000000 Before making any changes, first summarize the changes in a table. Then proceed with the updates. Do not modify demo copy or voiceovers.
A few things to like about this prompt: it's explicit about scope (only touching design and structure), and it asks for a summary before changes are made, which is useful when you want a quick review before execution. When it's done, you can review the response, let it make the updates, and then jump into the Supademo editor for a fast visual QA pass.
Use case 2: Personalize step text and voiceovers
The second example is rewriting step text in your brand tone and then generating concise voiceovers to match. Here's the prompt:
Example prompt:
I want you to review this Supademo: [INSERT DEMO ID OR URL] My goal is to improve the messaging so each step is more benefits-focused and aligned to the outcome we want for the viewer. Use this context when rewriting: - We want to position the product more clearly around business value and user outcomes without it sounding unnatural - The copy should feel clear, confident, and concise - Avoid generic filler language - This demo is specifically about our new RouteHub feature, for which we have docs here: <docs URL> For each step: 1. Propose improved on-screen text that is more benefits-focused 2. Propose a corresponding voiceover that is a shorter, focused variation of that text using 'Richard Yu' as the voice 3. Skip any chapter steps that are present Important: Do not change the text yet - output the proposed changes in a table before completing.
This is one of the best MCP workflows because it combines speed with guardrails. Instead of directly rewriting everything, you're asking Claude to first act like an editor. It looks at every step, proposes stronger copy, creates a shorter voiceover version for each one, and shows the full set in a table. That gives you a clean checkpoint before anything changes.
Once the table is generated, scan for three things: does the copy actually feel more benefits-focused? Do the voiceovers sound shorter and tighter than the text? And are there places where linking out to product updates or docs would improve the experience?
If the answer is yes, follow up with something simple like:
"Apply the proposed on-screen text and voiceovers for steps 1 through 12, but change steps 13 through 15 to be slightly more outcomes focused."
This is a great example of MCP complementing the built-in editor rather than replacing it. The MCP helps you get to a strong first draft quickly, then the editor is where you can fine-tune edge cases and do final QA.
Use case 3: Generate personalized tracking links at scale
The last example is one of the highest-leverage workflows for teams doing outbound sales, ABM, or customer follow-ups. Upload a mock spreadsheet of leads and ask Claude to generate tracking links for each one. Even better, you can add dynamic variables in the demo — like name, company, and role — which the MCP will personalize for each lead.
Example prompt:
I am uploading a spreadsheet of leads for this Supademo: [INSERT DEMO ID OR URL] Generate a unique tracking link for each lead in the spreadsheet and add them to a new column so we can share one core demo at scale while personalizing it per viewer. Make all the links expire on May 15th, 2026. Do not rewrite demo content in this workflow. Only focus on link generation and variable mapping.
The MCP generates several tracking links from one demo, without the need to duplicate it for each prospect. You keep one core asset, personalize the viewing experience through variables, and still get a unique tracking link for each recipient. Then take the CSV and use it for prospecting, sales, or in another tool's MCP.
When to use MCP vs. the built-in editor
Use MCP when...
- Making repetitive changes across many steps
- Rewriting copy or voiceovers in bulk
- Generating tracking links at scale
- The change is easier to describe than click through
Use the editor when...
- Fine-tuning specific visual details
- Doing final QA after MCP changes
- Pixel-perfect positioning of hotspots
- Making one-off, precise visual adjustments
Wrapping up
The big takeaway is this: MCP is not about replacing the editor. It's about speeding up the kinds of work that are repetitive, broad, and easier to describe than to click through manually. Use it when you know the outcome you want. And use the built-in editor when you want fine control and final polish.
If you're just getting started, try one of the three prompts from this video on a non-critical demo first, review the output carefully, and then build toward more repeatable workflows from there.