The Best GPT-4 Apps (so far)

Open AI announced a ton of new capabilities and products at their 2023 OpenAI Dev Day, and we are already seeing builders create new AI-powered apps using vision, longer context length inputs, agents, better AI models, and more.

If you’re looking for inspiration on what to build next or what the future looks like for AI apps, check out this list, which includes a round-up of the latest AI demos, along with inspiration on how you might be able to extend each of the ideas.

Here are some of the coolest apps we’ve seen powered by GPT-4 so far:

(PS: Have more AI apps to add to this list? Shoot an email to daniel - at - plusdocs.com, and let me know what we are missing! We’d love to feature more indie builders and early stage startups!)

{toc}

Best apps using the GPT-4-Vision API

Turn Mockups into HTML

@tldraw turned their online whiteboard into a magic designer and coder that can convert simple mockups into actual HTML code written with Tailwind.

It uses the GPT-4-Vision API to read the mockup and then convert it into an HTML file. It can even read diagrams to understand state and create more complex logical apps.

Link: https://makereal.tldraw.com/ (requires an Open AI API key)

David Attenborough Life Narrator

@charliebholtz built an app using GPT-4-Vision and the Eleven Labs text to speech API that uploads stills from your webcam to GPT-4. OpenAI then creates a narration in the style of David Attenborough, which is converted to speech to sound like David Attenborough is narrating your webcam.

It's actually quite funny. Worth checking it out!

Link: https://github.com/cbh123/narrator

AI Sports Announcer

‍

@geepytee created an AI sports narrator by using the OpenAI Cookbook for passing in still frames from a video and asking GPT-4V to create a ‘voiceover script in the style of a super excited brazilian sports narrator.’ 

From there, they used a text to speech tool to add the voiceover to the video, and turned it into an audio narration for the video file.

Ideas for where to bring this idea next? Add multiple voiceover tracks for a play-by-play commentator and a color commentator. Focus on a specific sport or team. Try narrator for less ‘sporty’ activities like chess or painting and see where that goes.

Landing Page Optimizer / Website roaster

@sethjkramer created this AI app that takes a URL, snapshots it as an image, and then uses the GPT-4V API to make suggestions to improve copy, design and layout, and overall structure. Customers can then use it to quickly improve their websites. Alternatively, @marcelpociot built a tool that will pull your website and roast it (and also return the audio!).

Ideas for where to bring this idea next? Ask the AI to find competitor websites and identify the best parts of each page. Fine tune the copy suggestions to match the site style. Link to other image sources to suggest image improvements.

Chat Webcam

We’ve seen a slew of apps where you can “Chat with a PDF” or quickly build a support bot using existing support materials. @skalskip92 and @benjaminDEKR used the GPT-4V API to build a chatbot that can “see” a user’s webcam and use that information in real-time.

Ideas for where to bring this idea next? Focus on a particular customer segment to build assistive tools for blind users. Build a scavenger hunt game for items around a house. Build Omegle v2 that matches people based on their environments.

AI Designer(s)

@sawyerhood was able to use the GPT-4-V API to ask GPT to convert an image of a low fidelity mockup into actual HTML code that would be usable in an app.

‍

Taking this idea to the next level, @jamescodez built a tool to use GPT-4 to create animations, then self-adjust the outputs to enhance the animations by itself!

Ideas for where to bring this idea next? Focus on a particular use case. Add in AI-generated images. Fine-tune outputs for a particular visual style (e.g., Apple announcement).

Best custom GPTs

Plus AI Presentation Maker

Plus AI is one of the best AI presentation makers that speeds up the process of making new presentations or editing existing presentations. They also have a custom GPT that will help you jot down the outline of a presentation, which can later be converted into a Google Slides or PowerPoint presentation using their AI tool.

Link to custom GPT: https://chat.openai.com/g/g-M1LP4l1Kf-plus-ai-presentation-maker

‍

Tweet optimizer GPT

@rowancheung downloaded his Twitter data from Twitter analytics, configured custom instructions, and then used it to create a GPT that can fine-tune posts and pick the best times for posting on X.

Ideas for where to bring this idea next? Automate the whole process so the human just has to click “Approve” and tweets are sent out. Recreate the same process for other forms of social media. Add images, videos, and other types of content.

Playlist builder

@brettunhandled added web browsing to his AI Spotify playlist builder to allow users to make queries for things such as “find the 2023 coachella lineup and make a playlist of the highlights.”

Ideas for where to bring this idea next? Build new types of playlists (e.g., TikTok playlists, YouTube playlists) and sort other types of information beyond songs (e.g., reading list, podcast list).

Best apps using the new OpenAI Assistants API

AI Travel Assistant

Hard to beat OpenAI’s own demo, but this travel assistant shows off how OpenAI-powered apps can build specialized assistants that are able to combine multiple features like function calling, web browsing, document uploads, and more. 

Ideas for where to bring this idea next? Focus on a particular type of trip or travel experience. Focus on hypercustomized experiences. Build integrations into other travel booking tools.

More coming soon!

Conclusion

Now is an incredible time for individuals and businesses to explore what is possible with AI. Seeing some of these early demos provides a hint into what should be possible across all types of applications in the next few years. Hopefully these glimpses serve as inspiration for what’s next for you!