🎯 Opportunity: Transform Teamwork with AI

Deeper Signals unlocks accurate, personal insights that help grow individuals, build teams and scale organizations. ****Part of their aim is to rapidly generate reports that address manager queries about individual and team challenges. Deeper Signals saw an opportunity to enhance teamwork with GenAI, augmenting an LLM with personality and team data to generate personalized, real-time reports based on core drivers.

🧩 Approach: Amazon Bedrock as Basis in a RAG World

To create and augment an LLM model, Deeper Signals turned to Loka and AWS. Loka utilized Amazon Bedrock with Anthropic’s Claude, together with Langchain and ChromaDB, to build the application’s foundational architecture. Loka used Retrieval Augmented Generation (RAG) to help the LLM grasp the Five Factor Model (FFM), Deeper Signal’s hierarchical categorization of personality traits into five basic dimensions, the 12CD Model, and their interrelation. Amazon Bedrock was ideal because they can switch models for different types of queries.

With RAG, Deeper Signals’ application can fetch context-specific data from external sources and make it available to an LLM when asking it to generate a response. Loka’s team was able to store a knowledge library containing a document collection about the FFM and 12CD Model and have their application fetch it for the LLM at generation time, reducing the likelihood of hallucinations. The result is a noticeable boost in the performance and accuracy of Deeper Signal’s GenAI application and reports.

🚀 Impact

• Automatically generated key insights about team members and suggestions to improve interactions.
• Eliminated several person hours per report.
• Leveraged GenAI to unlock a new service offering within Deeper Signals’ product.
• Gave managers a new ally in navigating individual challenges and scenarios, grounded in validated personality models.

🔑 Key Takeaways

• Amazon Bedrock was pivotal in enhancing product quality through sound model experimentation.
• The RAG-based solution outperformed foundational models, even though the information used to augment the model is publicly available in some form.
• Deeper Signals successfully overcame obstacles and integrated generative AI into their product.