5 Ways to Unlock AI Potential from Dreamforce 2024 Insights

AI is fundamentally reshaping how we work. From boosting productivity to driving efficiency and scalability, AI agents are redefining what’s possible in today’s business landscape. Yet behind these AI-driven processes lies a complex ecosystem of data sources, workflows, and legacy systems that must work together to realize AI potential. 

To tackle these challenges, Dreamforce introduced Agentforce — a complete AI system for enterprises that integrates data, automation, and humans in one solution. It’s a comprehensive agentic platform with everything you need to scale autonomous AI agents throughout your enterprise with low-code solutions. 

Here are the top takeaways from Dreamforce to help you create AI agents that can transform business operations and software development.

Takeaway #1: Having an agentic platform is key to building AI agents that exceed expectations

With CEOs prioritizing AI, it’s clear they recognize its potential to solve problems and improve experiences for employees, partners, and customers alike. 

For many teams, realizing this potential is a daunting challenge. A majority of IT leaders (88%) say they can’t meet AI demands safely — but safety is only one of the many roadblocks to AI. Legacy issues like disconnected data and workflows, tech debt, IT backlogs, and limited resources become even more pronounced with AI, creating a delivery gap between ambition and execution. 

The solution? A fully integrated, agentic platform that connects data, AI, automation, and humans into one unified system. With this approach, organizations can build flexible, secure AI agents and apps that deliver meaningful results.

The Platform keynote brought this to life with a live demo, showcasing how an Agentforce agent  was built on the spot to streamline casting for a hypothetical sports reality TV show, “Hooping with the Stars.” 

By seamlessly integrating past casting data, audition videos, and applicant details, the Agentforce agent provided actionable insights that helped talent managers make quicker, more informed decisions. It was a vivid demonstration of how an agentic platform makes it easier to build dynamic, reliable AI agents — transforming challenges into opportunities.

The Agentforce keynote highlighted a fundamental driver behind AI innovation: the desire to get work done. 

AI agents are a significant leap forward from traditional chatbots and copilots. While chatbots assist or answer questions, AI agents go further. They have reasoning engines that allow them to make decisions and take action. This capability has immense potential for enhancing customer experiences. AI agents can help employees eliminate repetitive tasks by taking action, allowing them to focus on higher-impact activities, especially those involving customers.

One compelling example comes from Wiley, a global leader in research and education. Wiley deployed Agentforce agents to manage routine inquiries in a pilot program, enabling employees to take on more complex cases. Agentforce agents outperformed Wiley’s previous chatbot within just weeks, improving case resolution rates by over 40%. 

Despite their sophistication, agents don’t have to be complicated to build. Salesforce designed Agentforce as a platform where “you can innovate without headaches,” as Gary Brandeleer, Senior Director of Product Management at Salesforce, said. With no-code and low-code tools like Agent Builder, Prompt Builder, and more, all on an integrated platform, anyone can build agents that understand your business and take action on your behalf.

Takeaway #3: Unifying structured and unstructured data enables more accurate and intelligent agents

The first-ever Data Cloud keynote spotlighted a key driver of many AI strategies — using data to enhance personalization and insights.

Data Cloud has already established itself as Salesforce’s fastest-growing organic innovation (and for good reason). Over the years, this platform has helped organizations address some of their biggest data challenges: eliminating data silos and making it easier to integrate data sources.

This past year marked a major evolution  for Data Cloud with the introduction of the Data Cloud Vector Database, which simplifies integrating unstructured data into workflows and AI agents. 

Structured data — think spreadsheets or tables arranged in columns and rows — is organized in consistent, standardized formats, making it easily sorted and analyzed. On the other hand, unstructured data — PDF documents, slide decks, and video and audio files — is far less accessible. With 80% of enterprise data being unstructured, Data Cloud’s recent expansion has massive potential for delivering 360-views of customers that can power more up-to-date, personalized experiences. 

A prime example is Mascoma Bank, which used Data Cloud to unlock data from 66 legacy systems. This allowed sales teams faster access to structured and unstructured data. Now, they can access company purchase history and deliver personalized recommendations, which has reduced manual data handling by 45 minutes per new loan.

Takeaway #4: The future of customer experience relies on automated systems built on a secure API architecture

In just a few words, the MuleSoft keynote kicked off with an undeniable truth — the success of AI relies on API-led integration. 

Beneath the surface of an organization, a complex web of systems is at work. On average, organizations have 991 applications, each with its own update schedule and security requirements. On top of that, 95% of IT leaders highlight integration challenges as a significant hurdle to successfully implementing AI.  

Organizations need an open and secure API architecture to build autonomous AI agents and apps, which can speak to and take action across the enterprise.

For over a decade, MuleSoft’s API-led approach has been key to unlocking data from any system, creating a foundation for autonomous agents to take action. But that’s just the beginning. With the introduction of MuleSoft Topic Center, developers can now package MuleSoft APIs and metadata into Topics to be deployed on Agentforce.

Topics combine APIs and metadata, teaching agents how to take action based on the data embedded in the APIs. Each API becomes a building block of actions agents can take across various systems. This was demonstrated live, where team members built an agent for Aston Martin, guiding customers through a VIP vehicle customization experience. The demo showed how APIs can create actionable agents, saving valuable time with just a few clicks.

Takeaway #5: Getting the correct data to the right people (and agents) is the key to better decision-making

The Tableau keynote summed up the hopes and dreams of decision makers when it comes to their data: the ability to interact and take action with their data quickly.

Oftentimes, users have to deal with the repetitive tasks involved in creating reports, building dashboards, configuring data sources, setting up visualizations, and manually updating metrics before they can get to the action.

A lot of time is spent clicking around, creating calculations from scratch, and figuring out what chart type to use. Now, with tools like Tableau Einstein, Tableau Pulse, and Tableau Agent — powered by Agentforce — analysts and non-technical users can visualize data by having a simple conversation using their preferred language: their own.

Using Tableau Agent, users can type in a question and ask the agent to visualize a piece of data. Then, Tableau Agent goes to work and builds the visual for them. What used to take hours for someone to develop, now takes only a few minutes.

Empower your autonomous agents with data, AI, automation, and humans 

Whether an organization is building AI agents for 24/7 customer service, creating dashboards to track metrics, or automating actions for sales teams, the success of these initiatives depends on seamlessly integrating data, AI, automation, and humans. 

That’s exactly what Salesforce Platform, Agentforce, Data Cloud, MuleSoft, and Tableau are helping customers achieve on their AI journeys.