InsightSphere

Sales teams spend hours gathering data and preparing proposals. InsightSphere automates the entire journey—from research to final strategy—using autonomous AWS AI Agents.

Comment

Inspiration

Every sales cycle starts with the same bottleneck: information overload and inconsistent preparation.
Sales teams spend hours collecting fragmented data from:

  • Company websites
  • Internal product documents
  • Prospect research reports

We wanted to transform this manual work into an intelligent, goal-driven AI conversation, where specialized agents could collaborate — just like a real sales team — but faster, smarter, and more consistent.

What it does

Sales Insight Generation is an AI-powered multi-agent system designed to autonomously research prospects, understand their business pain points, and generate customized technical solutions and sales strategies.

How we built it

We designed a multi-agent architecture orchestrated through AWS Strands Agents, with reasoning and knowledge retrieval powered by AI Agents that consume structured data directly from Amazon S3.

1. Knowledge Base Creation

  • Company website → processed by Research Agent → stored as structured data in Amazon S3
  • Technical docs / case studies → processed and uploaded to S3 for direct agent access
  • Prospect website → filtered and stored in a Prospect dataset (S3 bucket)

2. Pain Point Analysis

  • Pain Point Analyst Agent uses LLMs + Comprehend to extract top business challenges from the Prospect dataset.

3. Solution & Pitch Generation

  • Technical Agent generates a tailored solution proposal using contextual data from S3.
  • Sales Agent reads company + prospect context and crafts a personalized sales pitch.

4. Agent-to-Agent Conversation

  • Prospect Agent challenges the pitch across 10 Q&A rounds.
  • Router Agent routes each query to the right expert (Sales or Technical).

5. Post-Conversation Analysis

  • Conversation Analyst Agent performs sentiment, objection, and keyword analysis.
  • Sales Strategy Agent generates a final sales playbook with winning strategies, key phrases, and action steps.

Challenges we ran into

  • Ensuring each agent used only the relevant company data was complex.

Accomplishments that we're proud of

  • Reduced research & preparation time from hours → minutes
  • Generated consistent, tailored sales strategies from dynamic data
  • Demonstrated measurable agent reasoning, autonomous collaboration, and real-time insight generation

What we learned

  • How to design goal-oriented, multi-agent architectures using AWS Strands and custom AI Agents
  • The importance of prompt consistency and state management in long agent conversations

What's next for InsightSphere

  • Integrate live web APIs for financial and market news updates
  • Add voice-enabled interface for real-time sales call assistance
  • Expand into multi-language prospect simulations
  • Integrate Bedrock Knowledge Base and AWS Q Business for enhanced reasoning and persistent memory in future iterations

Built With

Share this project:

Updates