EchoSphere is basically a Reddit clone and has ~5000 LOC. We created a poll on Reddit to see what kind of an app people would want to see that would demonstrate capabilities of AI coding agents. The results were very consistent and it did lead us to focus more on expanding GPT Pilot to create apps with more features rather than smaller apps with more complexity. Currently, this is an app with most features that we built internally.

• Time spent to create this: 60 hours
• Github repo
• Video overview coming soon

Code Chat is one fun project that we made up as an example to showcase. The idea is that you can use it to ask the LLM questions about your codebase. You paste in the link to a public Github repo; it scrapes the code from Github, chunks it, sends it to the LLM for analysis that creates a description about what the code does and saves the descriptions into the database. Then, you can ask the app a question about the codebase and the codebase shows you the response. In the demo, we’re using GPT-3.5.

• Time spent to create this: 7 hours
• Github repo
• Video overview

This is a platform that enables users to do A/B testing for websites, allowing them to experiment with different versions of their web pages to determine which one performs better in terms of user engagement and conversions. For each user, it creates a JS snippet that the user adds to their website. Then, whenever the user creates a new test and picks which HTML elements they want to test, GPTOptimizely will inject A and B versions of the elements to half of the users. Then, it will track a click event on each version and display the results in the dashboard.

• Time spent to create this: ~3 days
• Github repo
• Video overview

OpenAI Playground on steroids – it enables you to load a conversation from a JSON array or enter it manually, run the conversation with the LLM an X number of times, save the conversation to the database, and load it back in. We actually use this app when we engineer a prompt and want to see how it behaves. The playground is not good enough because it’s time consuming to repetitively rerun a single prompt. Rather, we choose how many times we want to run it and let the app run the prompt repeatedly. Once it’s finished, we can go into analyzing the results. This might be useful for people who are building an AI app and need to optimize their prompts.

• Time spent to create this: ~2 days
• Github repo
• Video overview

I’ve been releasing open-source projects for years now, and I always wanted to check how fast is my Github repo growing by comparing it to some other successful repositories on https://star-history.com/. The problem is that on Star History, I’m not able to zoom into the graph, so a new repo that has 1000 stars cannot be compared with a big repo that has 50.000 because you can’t see how the bigger repo does in it’s beginning. So, I asked GPT Pilot to build this functionality. It scrapes Github repos for stargazers, saves them into the database, plots them on a graph, and enables the graph to be zoomed in and out.

• Time spent to create this: 6 hours
• Github repo
• Overview video link

This is also one app we use internally to analyze a local SQLite database. It pulls the data from the database in a format that’s very specific to the GPT Pilot database structure, but it can easily be modified to fit other structures. It reads and uploads your SQLite database, splits the rows by the specific field, unpacks the rows into values, loads the LLM conversation data into a form, and enables you to simply change one of the messages and submit the LLM request to GPT-4 to see how the results will look like. This way, we can analyze the conversations that GPT Pilot’s agents have with the LLM and easily try what would happen if the prompts were different.

• Time spent to create this: ~2 days
• Github repo
• Video overview

Vanish Talk is a sample app created to showcase real time messaging features but so it's not too easy, other features are added. It has the ability to authenticate the user, send verification email, create real time chatting room with or without a password, join any open room, and has a cronjob running that deletes all the messages every hour.

• Time spent to create this: 4 hours
• Github repo
• Video overview