Sessions

Explore sessions. Plan your Dreamforce.

Browse the full session catalog by role, industry, and product to design an agenda that works for you. Reserve your seat for hands-on trainings, certification exams, roundtables and more.

Enthusiastic conference attendees smiling and laughing during a live presentation session, decorated with festive star and emoji illustrations.

More sessions are being added. Check back often.

You can learn more about our session formats here.
Explore All Sessions

Tip: Use " " around your search terms to find an exact phrase.

Community Speaker

Reserved Seating

Live Stream

Video On Demand

Back to Session CatalogBack to Session Catalog
How Salesforce Builds Smarter Agents with Data 360
Your agent is only as smart as your data. Learn how Data 360 & Agentforce work together to activate trusted knowledge for smarter self-service, and the real lessons from over 4 million conversations.

, Lead GenAI Strategy Program Manager, Salesforce

, Senior Product Manager, Salesforce

Product: Data 360
Role: IT Business Leader
Topic: Salesforce on Salesforce
Session Level: Intermediate
Session Format: Breakout
  • Tuesday, Sep 152:00 PM - 2:40 PM PDT
    (9:00 PM - 9:40 PM GMT)
    Moscone West, L3, Room 3020
    FIRST COME FIRST SEATED
Session Summary: The session highlighted how Salesforce utilizes Data Cloud to enhance the capabilities of Casey the Help Agent, improving customer support interactions through data-driven innovations. With over 60 million annual site visits and a 20% year-over-year increase, the team identified challenges in managing vast amounts of information and launched Agentforce, which has already handled over 5 million conversations. The speakers emphasized the importance of high-quality data, rigorous evaluations, and iterative improvements to ensure the AI agent provides accurate, relevant, and complete responses. Through a structured agentic development lifecycle and a combination of quantitative and qualitative evaluations, Salesforce achieved a 93% resolution rate for Informatica-related queries by optimizing content ingestion and leveraging tools in Data Cloud and Agentforce effectively.
Key Takeaways:

• High-quality content and data are foundational to the success of AI agents, as poor data quality directly impacts agent performance and answer quality.

• Agents are non-deterministic systems, requiring iterative experimentation and evaluation, rather than a traditional software development approach, to achieve reliable outcomes.

• Evaluations require a human-in-the-loop. Defining benchmarks for answer and service quality, such as golden data sets and acceptance criteria, is critical to evaluating and improving the performance of AI agents.

Session Summary Feedback: Report a Concern

I go to Dreamforce because innovation is transferable across industries, and I can connect with a community that stretches across cities, time zones, and years.

Kathy Lueckeman

Chair, Education Trailblazers Association