Ep 632: ChatGPT Apps: 3 Hands-on approaches to save time today
🎯 Summary
Podcast Summary: Ep 632: ChatGPT Apps: 3 Hands-on approaches to save time today
This episode of the Everyday AI Show, hosted by Jordan Wilson, focuses on the practical utility and future implications of OpenAI’s newly released ChatGPT Apps, which embed native third-party experiences directly within the ChatGPT interface. The core narrative centers on eliminating “file finishing friction”—the time lost copying, pasting, and reformatting between ChatGPT and external applications—by integrating services like Spotify, Canva, and Figma directly into the chat UI.
The host argues that while the initial rollout is imperfect (only seven apps available, and current exclusion of Business/Enterprise users), this shift toward interacting with websites via a chat UI represents the future of work, similar to the success of mobile app stores.
1. Focus Area
The discussion centers on the practical application of ChatGPT Apps (native third-party integrations within ChatGPT) to enhance productivity. Specific technologies discussed include multimodal interaction (text and computer vision), the difference between Apps, Connectors, and Custom GPTs, and live demonstrations using the Canva and Figma apps.
2. Key Technical Insights
- Native GUI Integration: ChatGPT Apps are distinct from previous attempts (Plugins, GPTs) because they offer a native graphical user interface (GUI) embedded within the chat window, allowing users to interact with external services directly via natural language.
- Multimodal Context Awareness: The apps leverage the LLM’s multimodal capabilities, meaning they can process visual inputs (like an uploaded mind map image) and apply that context to the task within the third-party application (e.g., recreating the map in Figma).
- Productivity Gains: OpenAI internally reported a 70% productivity increase when using these integrated apps compared to the traditional workflow of switching between windows and manually transferring data.
3. Business/Investment Angle
- Ecosystem Play: OpenAI is aggressively pursuing an ecosystem model, similar to Apple’s App Store, evidenced by the introduction of a new protocol to handle in-chat financial transactions and checkouts.
- Revenue Potential: The host predicts this could become a multi-billion dollar endeavor within the first year if successful, driven by potential transaction fees (speculated to be 20-30%) on commerce conducted within ChatGPT.
- Market Validation: The partnership with Walmart for in-chat shopping signals a strong push toward embedding commerce directly into the AI interaction layer.
4. Notable Companies/People
- OpenAI: The developer of the ChatGPT Apps, aiming to close the gap between AI output and final execution.
- Jordan Wilson (Host): Provides the analysis and live demonstrations.
- Mentioned Apps: Spotify, Zillow, Canva, Figma, Expedia, Booking.com, and Coursera.
- NotebookLM: Mentioned as the tool used to generate the source material (a mind map) for one of the demos.
5. Future Implications
The conversation strongly suggests that the industry is moving toward a “chat UI as a browser” model, where users will increasingly manage complex tasks across multiple services without leaving the conversational interface. This will necessitate that all major software providers integrate deeply with leading LLM platforms to remain relevant.
6. Target Audience
This episode is highly valuable for AI Practitioners, Business Leaders, and Power Users who already rely heavily on ChatGPT and are looking for immediate, actionable strategies to integrate these new features into their daily workflows to boost efficiency.
Comprehensive Narrative Summary
The episode begins by identifying the “file finishing friction” that plagues current AI usage—the tedious process of moving finalized AI outputs into their final destination applications. Jordan Wilson introduces ChatGPT Apps as OpenAI’s solution to this problem, positioning them as a significant step toward the future of work where users interact with external websites via a chat UI.
Wilson differentiates Apps from previous features: Connectors bring in enterprise data, and Custom GPTs are personalized versions of the model, whereas Apps are native, embedded GUIs for third-party services (like Canva or Coursera). He notes that this is OpenAI’s third attempt to build an ecosystem, following the failure of Plugins and the relative stagnation of GPTs.
A key point of discussion is the surprising availability of Apps to free users. Wilson explains this is strategic: OpenAI wants maximum adoption to test and refine the new in-chat checkout protocol, paving the way for future revenue generation through transaction fees, mirroring the success of mobile app stores.
The bulk of the episode is dedicated to three hands-on approaches demonstrating immediate time savings:
- Bringing in Images/Elements (Multimodality): A demo using the Figma app showed how uploading a mind map image (generated in NotebookLM) allowed ChatGPT to analyze the visual structure via computer vision and recreate the entire diagram vertically with new color schemes, saving hours of manual design work.
- Enriching Starting Points with Connectors: This was partially demonstrated in the Coursera example, where prior research context was established.
- Giving Personal/Professional Context: The host set a specific persona (expert in AI, learning neural networks, needing 90s basketball analogies). When asking for a Coursera course on neural networks, ChatGPT correctly surfaced the Coursera app, embedded the course player, and—crucially—retained the established context when asked follow-up questions about the course content, even integrating the requested niche analogy.
Wilson concludes that while the current selection of seven apps is limited, users must start practicing with them now, as this integrated, context-aware interaction model
🏢 Companies Mentioned
đź’¬ Key Insights
"I'm constantly doing research in ChatGPT, deep research, Perplexity, Gemini deep research, all these things, and almost every single time that I'm doing a very specific deep research on something, my content is some of the content that comes up. It's a problem for me. I forget a time."
"It analyzed the image with computer vision that I uploaded from the NotebookLM mind map, and then it recreated it, but not only did it recreate it node per node, and now it is going to be editable inside Figma, which I'll show you, but it also changed it pretty drastically."
"You have to think of different modalities when you think of context because it's not just, 'Oh, I share some text, and then that text can be taken into account when I'm interacting with an app.' Right? Large language models are multimodal by default, input and output."
"According to OpenAI and some of their internal testing, overall, they saw a 70% productivity increase working in these apps versus if you were doing the thing I was talking about earlier, the transfer tax, the format fallout..."
"They want people using ChatGPT as if it was a browser because ultimately they will be collecting money on different types of transactions that happen within ChatGPT."
"We are going to be interacting with websites in a chat UI. That's what apps are all about from OpenAI."