Skip to main content

Salesforce's Rahul Auradkar on Unified Data Engines and Agentic AI Context

Super Data Science: ML & AI Podcast with Jon KrohnJanuary 17, 202624 min9,505 views
41 connections·40 entities in this video→

The Need for Context in AI Agents

  • πŸ’‘ AI models often perform poorly because they lack the necessary context, leading to "stupid" actions.
  • 🎯 Salesforce's unified data engine aims to provide this crucial context, making AI agents more intelligent and effective.

Salesforce's Unified Data Engine Components

  • 🧩 Data 360 is Salesforce's offering, focusing on customer data platforms (CDPs) to unlock, harmonize, and activate data for insights and customer experiences.
  • πŸ“Š Tableau provides connect-and-explore analytics, now enhanced with a semantic layer for actionable and agentic analytics, enabling natural language queries.
  • πŸ”— Mulesoft acts as an integration platform (iPaaS) for app-to-app integration and API management, extended to govern and orchestrate actions for agents through an "agent fabric."

The Informatica Acquisition and its Impact

  • 🀝 The acquisition of Informatica significantly augments Salesforce's data foundation, bringing leadership in data quality, data cataloging, data integration, and governance.
  • πŸ”‘ Informatica's enterprise data catalog provides a superset of metadata, while its ETL tooling enhances data integration capabilities.
  • πŸš€ This integration is key to building the AI foundations needed for data fluidity and providing trusted context to AI agents.

Bridging the Context Gap

  • ⚠️ Enterprises are often data-rich but context-poor, leading AI models to provide disconnected or irrelevant responses.
  • πŸš— An example highlights how a sales system, marketing system, and customer service system lacking context would fail to recognize a car purchase, leading to inappropriate advertisements instead of relevant offers like insurance or accessories.
  • βœ… Grounding AI agents with unified, trusted context leads to more delightful and relevant customer experiences.

Upskilling for Data Science Teams

  • πŸ› οΈ Salesforce emphasizes no-code and low-code tools to allow users to focus on their domain expertise, reducing the need for deep technical upskilling in every area.
  • πŸ“š The company supports learning through its Trailhead platform, offering lessons and certifications for its community and partners.
  • πŸš€ The goal is to enable customers to excel in their domains by leveraging the integrated capabilities of the unified data platform.
Knowledge graph40 entities Β· 41 connections

How they connect

An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.

Hover Β· drag to explore
40 entities
Chapters10 moments

Key Moments

Transcript89 segments

Full Transcript

Topics15 themes

What’s Discussed

Unified Data EngineAgentic AIData ContextSalesforceData 360TableauMulesoftInformaticaData QualityData CatalogData IntegrationETLCustomer Data Platform (CDP)Semantic LayerAPI Management
Smart Objects40 Β· 41 links
CompaniesΒ· 4
PeopleΒ· 2
ProductsΒ· 11
ConceptsΒ· 23