TP ICAP Boosts CRM Insights with Amazon Bedrock’s AI Power
TP ICAP’s Innovation Lab developed ClientIQ, an enterprise-grade solution leveraging Amazon Bedrock to transform vast amounts of CRM meeting notes into real-time, AI-powered insights. Facing the challenge of underutilized qualitative data within tens of thousands of Salesforce records, ClientIQ enables business users to interact with this data using natural language queries, drastically reducing manual analysis time from hours to seconds.
Key features include a conversational interface supporting plain English questions like “How can we improve our relationship with customers?”, query refinement, temporal filtering, and direct links to source Salesforce records for full traceability. ClientIQ intelligently routes user requests: unstructured meeting notes are processed via a Retrieval Augmented Generation (RAG) workflow using Amazon Bedrock Knowledge Bases, while analytical queries on structured data are handled by a text-to-SQL generation workflow. The solution respects existing granular permission boundaries, ensuring users only access authorized data through a robust security framework integrating Okta group claims with data metadata.
Technically, ClientIQ is built on Amazon Bedrock, utilizing various foundation models (FMs) like Anthropic’s Claude 3.5 Sonnet for classification and Amazon Nova Pro for text-to-SQL generation and topic tagging. Amazon Bedrock Knowledge Bases, with Amazon Titan v1 embeddings and Amazon OpenSearch Serverless as the vector store, power the RAG component. This employs a sophisticated hybrid search approach, combining metadata filtering with semantic search for enhanced retrieval accuracy. Structured data is queried via Amazon Athena from an Amazon S3 data lake, populated by a custom, daily-running Salesforce connector.
Benefits include a reported 75% reduction in research time and significantly improved insight quality, providing more comprehensive and contextual information. Automated evaluation, using Amazon Bedrock Evaluations integrated into the CI/CD pipeline, ensures consistent performance, quality, and responsible AI metrics. The primary target audience is TP ICAP’s internal stakeholders needing rapid, comprehensive insights from CRM data, with plans to evolve ClientIQ into a broader intelligent virtual assistant across multiple enterprise systems. This demonstrates how organizations can rapidly deploy production-ready AI solutions to unlock significant business value in weeks.
TP ICAP’s implementation demonstrates how ai automation crm solutions can transform financial services by delivering enhanced customer relationship management capabilities.
This advancement highlights the growing trend of chatgpt automation crm integration across financial services to enhance client relationship management capabilities.
