Bot Lens

Inspiration

The inspiration for Bot Lens came from the rapid growth and adoption of chatbots and conversational AI platforms. We recognized the need for an analytics solution that would help developers understand user interactions with their chatbot plugins, enabling them to iterate faster and optimize the overall user experience. By drawing from the successes of analytics platforms like Google Analytics, Mixpanel, and Flurry Analytics, we aimed to create a robust tool that could bring similar insights and capabilities to the world of chatbot development.

What it does

Bot Lens is an analytics platform specifically designed for chatbot plugins, such as those developed for ChatGPT. It provides developers with valuable insights into user behavior, interaction patterns, and usage trends, helping them make informed decisions about the design and functionality of their chatbot plugins. Key features of Bot Lens include:

Analysis of user interactions: Bot Lens tracks user engagement with various parts of the plugin, identifying which features are being used the most and which ones may need improvement. Conversation metrics: The platform monitors the number of conversations the chatbots are having, giving developers an idea of their chatbot's popularity and usage patterns. Endpoint usage analysis: Bot Lens helps developers identify the most frequently used endpoints, enabling them to optimize their plugin's performance and user experience. Geographical usage data: The platform provides information on where in the US (or other countries) the plugin is being used, allowing developers to target specific regions or adapt their plugins to better suit local user needs.

How we built it

We built Bot Lens to be reliable and easy to use. For collecting data from plugins, we implemented a lightweight Python SDK developers can easily integrate into their code. This SDK can pick up relevant data on user interactions, conversations, endpoint usage, and other events developers choose. The SDK then forwards data to our backend.

We set up the backend using FastAPI and Apache Druid to process and analyze the data in realtime. The backend ETLs and streams data from the SDK into our database, and responds to queries from our React frontend for visualization.

Bot Lens was built using a combination of modern web development technologies and analytics tools. To create a scalable and reliable solution, we leveraged cloud-based infrastructure and serverless computing for processing and storing data. We utilized popular web frameworks and libraries for building the front-end user interface, providing a seamless and intuitive experience for developers.

To collect data from chatbot plugins, we implemented a lightweight Python SDK that can be easily integrated into any ChatGPT plugin. This SDK collects relevant data on user interactions, conversations, endpoint usage, and geographical information, and sends it to our servers for processing and analysis.

On the backend, we used a combination of big data processing tools like Apache Druid and machine learning algorithms to derive actionable insights from the collected data. The processed information is then visualized through a series of interactive dashboards and reports, providing developers with a comprehensive view of their plugin's performance and user engagement.

Challenges we ran into

We ran into difficulty hosting the platform manifest on our own site & used replit to host our plugin instead.

Accomplishments that we're proud of

We are really proud of how comprehensive the solution we built is - we built out a python SDK, OLAP engine, Admin dashboard, & two chatGPT plugins.

What we learned

The plugin ecosystem is about to explode & analytics are the next major step to help companies make plugins.

What's next for Bot Lens

Building out more charts & better real-time support. We want to help people make these insights actionable.

Built With

• imply
• openai
• python
• replit