Sales ops · Production

TP ICAP builds ClientIQ with Amazon Bedrock to cut CRM research time by 75%

The problem

TP ICAP had accumulated tens of thousands of vendor meeting notes in their CRM that were being underutilized, with business users spending hours manually searching through records knowing the information existed but unable to efficiently locate it.

Workflow diagram · grounded in source
1
User submits natural language query
trigger
“users can interact with their Salesforce meeting data through natural language queries”
2
LLM analyzes query intent
ai_action
“It uses a large language model (LLM) to analyze each user query to determine the optimal processing path”
3
Route to RAG or SQL workflow
routing
“It routes requests to one of two workflows”
4
RAG hybrid search retrieval
ai_action
“ClientIQ uses hybrid search to first filter documents based on their metadata and then perform semantic search within the filtered results”
5
SQL generation for analytical queries
ai_action
“Amazon Nova Pro for text-to-SQL generation”
6
Natural language response delivered
output
“It then generates the responses in natural language”
7
Daily Salesforce data ingestion
integration
“The connector, which employs Salesforce Object Query Language (SOQL) queries to retrieve the data, runs daily and has proven to be fast and reliable”
8
Topic tagging with Amazon Nova Pro
ai_action
“automatically tagged with relevant topics from a predefined list, using Amazon Nova Pro”
9
Embedding and vector index update
integration
“each CSV file is converted into embeddings using Amazon Titan v1 and indexed in the vector store, Amazon OpenSearch Serverless”
10
Automated RAG evaluation in CI/CD
feedback_loop
“they integrated RAG evaluation directly into their continuous integration and continuous delivery (CI/CD) pipeline, so every deployment automatically validates that changes don't degrade response quality”
Reported outcome

Following the initial launch with 20 users, ClientIQ delivered a 75% reduction in time spent on research tasks, and stakeholders reported an improvement in insight quality with more comprehensive and contextual information being surfaced.

Reported metrics
Time spent on research tasks75%
Insight qualityimprovement in insight quality
Manual analysis time transformedhours of manual analysis into seconds
Time to build production solutionweeks rather than months
Show all 6 reported metrics
time spent on research tasks75%
insight qualityimprovement in insight quality
manual analysis time transformedhours of manual analysis into seconds
time to build production solutionweeks rather than months
vendor meeting notes in CRMtens of thousands
initial launch user count20
Reported stack
Amazon BedrockAmazon Bedrock Knowledge BasesAmazon Bedrock EvaluationsAmazon Titan v1Amazon OpenSearch ServerlessAmazon AthenaAWS GlueSalesforceReactOktaRAGSOQL
Source
https://aws.amazon.com/blogs/machine-learning/how-tp-icap-transformed-crm-data-into-real-time-insights-with-amazon-bedrock?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Following the initial launch with 20 users, ClientIQ delivered a 75% reduction in time spent on research tasks, and stakeholders reported an improvement in insight quality with more comprehensive and contextual inform…

What tools did this team use?

Amazon Bedrock, Amazon Bedrock Knowledge Bases, Amazon Bedrock Evaluations, Amazon Titan v1, Amazon OpenSearch Serverless, Amazon Athena, AWS Glue, Salesforce, React, Okta.

What results were reported?

Time spent on research tasks: 75%; Insight quality: improvement in insight quality; Manual analysis time transformed: hours of manual analysis into seconds; Time to build production solution: weeks rather than months (source-reported, not independently verified).

How is this sales ops AI workflow structured?

User submits natural language query → LLM analyzes query intent → Route to RAG or SQL workflow → RAG hybrid search retrieval → SQL generation for analytical queries → Natural language response delivered → Daily Salesforce data ingestion → Topic tagging with Amazon Nova Pro → Embedding and vector index update → Automated RAG evaluation in CI/CD.

Related sales ops cases