Agentforce: Configure Powerful GenAI Chatbot Experiences

We got our hands on Agentforce. Learn how this new no code GenAI chatbot can quickly be built to deliver incredible results. 

Generative AI can deliver great business results in areas that really matter. But building a real enterprise solution is challenging because many specialist AI and engineering skills are necessary to create the type of grounded, secure experience required by enterprise standards.

With the launch of Agentforce, this is no longer the case. 

Our team of AI and Salesforce experts has had a chance to get our hands on Salesforce’s new chatbot and I couldn’t be more excited about what we’ve been able to build for our customers.

GenAI Made Easy

We’ve been working in GenAI since the launch of ChatGPT nearly 2 years ago, and we’ve used OpenAI’s APIs to create myriad chatbots. Around a year ago, we created an AI prototype that could power a chatbot for a major airline like our partners at Spirit Airlines, which our world class engineering team built using MuleSoft and OpenAI.

Previously, a GenAI chatbot like Agentforce would’ve required a hefty engineering budget and skilled staff to build, test, and roll out. No more. The demo video above could have been built using Agentforce, Data Cloud, MuleSoft, and a little Salesforce configuration. 

With Agentforce, you can easily configure your chatbot to manage specific topics using natural language prompts similar to those you’d use with ChatGPT, allowing for a guided conversation tailored to your needs. By leveraging Salesforce automations like Apex and Flow, you can drive actions based on Agentforce’s interactions with customers. These automations are versatile and not limited to the chatbot; any existing workflows you’ve created are fully integrated and ready to use.

Moreover, Agentforce can be set up to perform nearly any task a human agent can within Salesforce. By utilizing powerful integration tools like MuleSoft, you can extend Agentforce’s capabilities beyond Salesforce, tapping into your entire IT infrastructure. This means you can harness data from various backend systems, enhancing your AI decision-making power and delivering a seamless experience for your customers.

We are already working with customers to deliver these chatbots quickly and iterate on their results. Let’s dive deeper into what makes next-generation customer service workflows with Agentforce so different.

Building Next-Generation Service Workflows with AI 

Great Generative AI capabilities all start with the human/AI interaction, and Agentforce is no different. While its powerful capabilities enable it to deflect even more cases than a traditional chatbot, some service interactions require a more human touch.

This is where Agentforce truly shines, taking advantage of Generative AI capabilities within Salesforce Service Cloud to support customer support representatives like never before.

The handoff to a human representative is truly remarkable. Conversation Catch-Up gives the customer support representative a stunningly accurate summary of the conversation that’s taken place between Agentforce and the customer, enabling the CSR to get up to speed quickly. Any details necessary are also available since case routing preserves the entire conversation, making all details readily available to support the customer quickly.

Einstein Service Replies give the CSR options for very quick responses to the customer’s inquiries, all grounded by Salesforce Knowledge. Gone are the days of having to search through a Knowledge Base manually to obtain the right information; we’ve fast-forwarded to a time where great responses are automatically generated from every piece of knowledge available, all in the voice of the brand.

When cases are wrapped up, Work Summaries leverage the same Generative AI capabilities to save the agent a lot of time. Cases are closed out by identifying the overall issue, resolution, and a summary of the chat that took place. Important yet time-consuming work that would’ve taken time away from servicing the customer is instead delegated to AI so that the CSR can focus on providing the best possible customer service.

Getting Started with Agentforce

Even though the Agentforce chatbot is technically capable of performing any action that a human customer service representative can because it has access to Flows and Apex, we’re not there yet. We’ve seen the most successful outcomes delivered by focusing on deflecting a few high-volume, easy-to-resolve case topics to save time from human agents and building from there. This type of approach enables Salesforce customers to fine-tune Agentforce to ensure it’s representing the brand the way it should before broadening its horizons to even more topics.

As expected, Agentforce comes with a robust set of test tooling and reporting capabilities geared towards ensuring that it’s operating to the highest standards. Automated test tools enable you to test Agentforce’s responses at scale to different phrases and wording, giving you confidence that the AI is well-aligned to its configured goals. Reporting capabilities allow administrators to hone in on any problem areas and quickly remediate them by fine-tuning Agentforce’s prompts.

In summary, customers can “evolve” Agentforce’s AI by piloting with a limited set of topics, performing a robust set of tests, and then monitoring and fine-tuning once live. This enables them to have the confidence that Agentforce is well-aligned to their goals and the voice of their brand and then expand the scope of its capabilities. And the scope is virtually limitless; since its actions are based on standard Salesforce capabilities like Flow and Apex, you can leverage all of the great work you’ve already done on the Salesforce platform to deliver exceptional customer experiences.

The Time Is Now

With GA a few weeks away, we are already working with several customers to prepare to rapidly deliver Agentforce.