Building Alice’s Brain: an AI Sales Rep that Learns Like a Human
summary
An AI SDR called Alice that learns by ingesting sellers’ PDFs, websites, calls and videos is built end to end: vendor parsing with Llama Parse, Firecrawl, Cloudglue, chunking into markdown, embeddings into Pinecone, and a deep-research agent (Leta) driving retrieval-augmented message generation. The system sends 50,000 outbound emails per day across ~300 customers, reduces manual library entry, and surfaces source chunks in a 3D knowledge-visualization—deployed and producing measurable outreach at scale.
AI-written summary · about
chapters (18)
- 0:00Introduction to Alice
- 1:06What is an SDR
- 3:53Old library system problems
- 4:55Knowledge base concept
- 5:38Knowledge base overview
- 7:45Pipeline steps overview
- 8:06Parsing process explained
- 9:50Vendor selection criteria
- 11:16Document parsing tools
- 12:40Chunking strategy
- 15:10Storage technologies
- 15:52Vector database choice
- 17:16Retrieval evolution
- 17:59Deep research agent
- 18:40Visualization purpose
- 20:23Project outcomes
- 20:44Lessons learned
- 21:47Future plans
related talks
22mFireside Chat: AI and Data-Driven Sales
22mAI- and Data-Driven Sales
27mBuilding AI applications: Successes and Lessons
30mRobots For Hire - How an A.I. Sales Rep Helps You Revolutionize Your Business
37mAlice's Adventures in Artificial Intelligence-land
22mFrom Alpha to Action: How AISDK5 Helped Us Build an AI-Native CRM
29mBuilding an A.I. Assistant that Speaks Human
36mAI Agents in Automation | 3 Real Production Cases with ROI