Salesforce, Databricks, AWS, & Madrona on the Data Innovations Powering Enterprise AI Agents

Salesforce, Databricks, AWS, & Madrona on the Data Innovations Powering Enterprise AI Agents

Oct 9 · 23 min

People

  • Ganapathy Krishnamoorthy Guest
  • Zaheera Valani Guest
  • Rahul Auradkar Guest

Listen on

Suggest an edit

Madrona Venture Partner Mark Nelson moderates a panel at the IA40 Summit with data leaders from @amazonwebservices , @Databricks , and @salesforce. They explain what context layers, knowledge graphs, and ontologies mean for enterprise AI agents, and why an AI strategy is a data strategy. They cover cost, accuracy, headless applications, and the database features that agents are changing. Rahul Auradkar, President & GM, Data 360, Salesforce Zaheera Valani, VP Engineering, Databricks Ganapathy (G2) Krishnamoorthy, VP, Database Services, AWS 00:00 – Why an AI Strategy Really Means a Data Strategy 01:15 – What Context Layers, Knowledge Graphs and Ontologies Mean for Enterprise AI 04:15 – Why Hand-Modeled Context Goes Stale and How Databricks Auto-Generates It 06:24 – Why AI Models Are "Corporate Stupid" and Expensive: Salesforce on Minimal Context 08:29 – How Wyndham Got a 300 Basis Point Conversion Lift From Graph-Based Context 09:55 – AI Accuracy With a Context Layer: 80% vs 50% Without 12:55 – Data Gravity Is a Thing of the Past: Salesforce's Zero Copy Network 13:28 – The Car Purchase Story That Shows What Data Silos Cost 14:24 – Why the Database Is the Ultimate Headless App 15:34 – How "Video Was Viewed" Meant Different Things Across Apps at Meta 18:06 – Headless Salesforce and Jason Lemkin's 9X Increase in Usage 20:11 – Dropping a .md Data Model Next to the Database for AI Agents

Salesforce IA40 Summit Mark Nelson Databricks AWS Madrona

Chapters

  1. 0:00 Why an AI Strategy Really Means a Data Strategy
  2. 1:15 What Context Layers, Knowledge Graphs and Ontologies Mean for Enterprise AI
  3. 4:15 Why Hand-Modeled Context Goes Stale and How Databricks Auto-Generates It
  4. 6:24 Why AI Models Are "Corporate Stupid" and Expensive: Salesforce on Minimal...
  5. 8:29 How Wyndham Got a 300 Basis Point Conversion Lift From Graph-Based Context
  6. 9:55 AI Accuracy With a Context Layer: 80% vs 50% Without
  7. 12:55 Data Gravity Is a Thing of the Past: Salesforce's Zero Copy Network
  8. 13:28 The Car Purchase Story That Shows What Data Silos Cost
  9. 14:24 Why the Database Is the Ultimate Headless App
  10. 15:34 How "Video Was Viewed" Meant Different Things Across Apps at Meta
  11. 18:06 Headless Salesforce and Jason Lemkin's 9X Increase in Usage
  12. 20:11 Dropping a .md Data Model Next to the Database for AI Agents