.webp){kind=logo}
Case Study
Animal Ethics. Senti Bot
Animal Ethics, an organization focused on promoting respect for non human animals through outreach, research, and education, wanted an AI assistant to make it easier to educate about using future technologies to mitigate harms in nature to animals, particularly on a large scale. The main challenge was designing an AI-driven chatbot that accurately reflected the philosophy of Animal Ethics, particularly within the constrained timeline of eight weeks.
Non-Profit
2023
Services
User, Business, and Technical discovery
Software architecture consulting
Agile development
Project management
Software design and development
UX/UI design
Technology Used
Jupiter Notebooks
BrowserStack
Figma
Azure DevOps
GO
Typescript
Java
Claude AI Chat
Open AI Embeddings
Client feedback
Our collaboration with Freeport Metrics was fantastic. They built a flexible and scalable solution that addressed our biggest problem -- overcoming pervasive biases that reflect common human misconceptions. The team took the time to understand our philosophy and came up with creative ways of testing. They made sure we were part of the process every step of the way, and got it all done in lightning time.
Co-founder at Animal Ethics
Challenges
Reflecting a complex philosophy
Animal Ethics upholds a unique philosophy that prioritizes sentient beings' welfare, sometimes above environmental considerations. The challenge involved creating an AI assistant that would precisely mirror their philosophy without appearing biased to people who might hold opposing views.
Short timeframes
The challenge was amplified by extremely strict timelines. We needed two weeks to demonstrate our capability to control biases and set up the necessary tools, followed by six weeks for AI chatbot development.
Solutions
Feedback loop process
We set up a feedback loop with Animal Ethics to gather crucial information and improve chatbot responses. This process included identifying biases, providing context, testing Q&A, gathering client feedback on responses, and making prompt improvements. Collaboration proved to be the key to quick deployment.
Microservices solution
Our solution consisted of several components: Open AI embeddings, a Claude chatbot, a database, a service for importing documents, and an API gateway. This model enables easy adaptation of the LLM provider, which is particularly beneficial in the ever-evolving AI landscape.
Results
Hire usImproved accuracy of AI chatbot responses
Through the implementation of the feedback loop, we were able to obtain a quick response from Animal Ethics, instrumental in driving improvements in the chatbot's accuracy. By reviewing the comments and refining the given prompts, we managed to boost the chatbot's response correctness rate to 70%.
User-friendly interface with access to key topics
The developed chatbot retrieves quotes from the database, constructs responses, and verifies them against our bias list before delivering the final answer. It covers ten topics and offers users an informative and engaging experience. The chatbot is accessible via a floating button on the website, linking to external sources, and presents the three most important topics alongside example questions that users could potentially ask the bot.
