Build an AI System, Not a Tool Collection

I’ve seen so many people just hoarding AI tools, thinking the next shiny object will magically solve all their productivity problems. It’s an easy trap to fall into, but it rarely works. I just stumbled upon an incredible video from an AI professional that completely flips this idea on its head. The mind behind it argues that what we really need isn’t more tools, but better systems.

This is such a sharp insight. Instead of endlessly swapping between the ‘best’ AI of the week, we should be building intelligent workflows that connect the strengths of different models. It’s about making them work together, not against each other.

Your Team of AI Specialists 🧠

The creator proposes a brilliant mental model: think of your AI tools as a specialized team. Each member has a distinct role, and your job is to be the project manager who assigns the right task to the right specialist. I was blown away by how simple and effective this framing is.

Here’s the team roster this industry pro laid out:

  • ChatGPT: The Generalist Assistant. Your go-to for a wide range of tasks and initial brainstorming.
  • Gemini: The Integrated Builder. Perfect for creating things, especially within the Google ecosystem.
  • Claude: The Visual Strategist. Excels at thoughtful analysis and creating structured, beautiful visuals.
  • Perplexity: The Specialized Researcher. Your expert for finding and citing real-time, accurate information from the web.
  • NotebookLM: The Knowledge Synthesizer. The master of making sense of your specific documents and sources.

Once you see them this way, the question changes from “Which AI is best?” to “Who on my team is right for this job?”

Here are some of the most powerful workflows the creator demonstrated:

📌 Supercharge Your Research and Analysis

This isn’t just about asking an AI to “research a topic.” The creator shows two sophisticated methods for getting much deeper, more accurate insights. The first is about building a “Project Brain” in ChatGPT. Instead of starting with a blank slate, you create a dedicated ChatGPT Project and upload background documents, like a brand overview. Then, you run your deep research inside that project. ChatGPT uses the context you provided to generate a far more relevant report. The final step is key: you upload that new research report back into the project, continuously training your project brain. This creates a self-improving knowledge base you can query for anything from buyer personas to brand voice guides.

The second method gives you maximum control. It combines Perplexity and NotebookLM. The expert starts on Perplexity, asking it to find the top 20 URLs for a specific topic (like competitor websites) and to only output the URLs. You then copy that clean list of links and import them all directly into NotebookLM. This is brilliant because you’ve hand-picked the sources, drastically reducing the chance of AI hallucinations. NotebookLM will only analyze the information you’ve fed it. The creator then adds a persona in the chat settings, like “a growth consultant specializing in market positioning,” to frame all of the AI’s responses.

💡 Create Strategic Visuals from Raw Data

Turning a 20-page report into a shareable, easy-to-understand visual is a common struggle. This innovator shared two workflows for this. For quick visualizations, she combines Perplexity and Gemini. She uses Perplexity’s specialized “Social” search to find Reddit discussions about user frustrations with a product category. After exporting that research as a PDF, she uploads it to Gemini and uses the Canvas mode. Because of Gemini’s large context window, it can process the entire report and generate beautiful, bite-sized infographics and charts that tell a story. A fantastic tip she shared was prompting Gemini to “add a download as image button” directly to the canvas, making it easy to save the visuals.

For more strategic dashboards, the go-to combination is Perplexity and Claude. The process involves finding credible PDF reports with Perplexity and uploading them to a new Claude Project. You can then ask Claude to create an integrated dashboard to visualize the insights, even providing a specific color palette. Here’s the game-changing part: you can then save that entire dashboard back into the project as a reusable asset. This means the next time you have a similar report, you can just upload it to the same project, and Claude will generate a new dashboard in the exact same strategic format.

✅ Turn Your Ideas into Tangible Assets

This is where research becomes reality. The person who shared it showcased a few ways to create ready-to-use assets. One of my favorite tricks is using Claude to create editable social media graphics. The secret is asking Claude to generate an “SVG artifact.” Unlike a flat image, an SVG is a vector file that is fully editable. You can take the output directly into a design tool like Figma and change every single element, such as colors, fonts, and layout, to match your brand. You can go from a list of takeaways from Perplexity to a polished, custom social media graphic in minutes.

For bigger projects, like a full training package, the workflow is even more powerful. The creator uses NotebookLM’s “Discover Source” feature to gather authoritative articles and videos on a topic. She then asks it to design a course curriculum, which she saves as a note. She generates content for Module 1, saves it as a new note, and then uses NotebookLM’s “Video Overview” feature to create an entire video lesson based only on that module’s content. She even has it generate a toolkit with exercises. That toolkit text is then copied into Gemini’s Canvas mode to build a full presentation deck to accompany the video lesson. This same process can be used to build a business landing page, going from market research in NotebookLM to a fully functional HTML prototype in Gemini.

These workflows really opened my eyes to what’s possible when you stop thinking about individual tools and start thinking about integrated systems. This is how you get real leverage from AI.

I highly recommend you check out the full video from this talented creator to see these workflows in action!

Scroll to Top