If you could free up more time to focus on tasks that you enjoy and add value, would you? How much of your time do you seem to spend in the weeds….replying to similar emails, writing documents, creating marketing plans, when you should be focused on leading your team, or operating at that more strategic level? Agentforce is the new Salesforce Artificial Intelligence Platform that may just allow you to do that, without taking on more headcount. Too good to be true? Read on to find out!
Agentforce was announced at Dreamforce 2024 by Marc Benioff, but it wasn’t a last minute thought, it was an overnight success that took years of planning and testing. Salesforce, with Marc at the helm, has always been recognised for its pioneering mindset, and although Artificial Intelligence has become somewhat of a buzzword recently, we have to look all the way back to 2016, when Salesforce first released Salesforce Einstein, to see how far ahead of the curve Salesforce has been.
What is Agentforce?
“Agentforce represents the Third Wave of AI—advancing beyond copilots to a new era of highly accurate, low-hallucination intelligent agents that actively drive customer success……..Agentforce offers a fully tailored, enterprise-ready platform designed for immediate impact and scalability. With advanced security features, compliance with industry standards, and unmatched flexibility. Our vision is bold: to empower one billion agents with Agentforce by the end of 2025. This is what AI is meant to be.” Marc Benioff, Dreamforce 2024
Whereas the AI that we currently experience can provide good responses to simple questions, it is not a proactive Agent that will take action without your prompts and input. Agentforce aims to address this by becoming a fully integrated solution that will proactively take action based on the guidance you give it.
By accessing data across multiple platforms, and harnessing the power of Data Cloud, Agentforce can review huge amounts of data to get insights into customers within Salesforce, and then build personalised actions based on that. Whereas Einstein AI delivered prompt-reply functionality, Agentforce can be left to interact with customers, and build plans autonomously.
Agentforce can be split into four key areas:
Tools
Agent Builder: creation and customisation of Agents – Topics, instructions, Actions (Prompts/Flows/Apex)
Agentforce Reasoning Engine: how Agentforce takes action – Retrieval Augmented Generation (RAG) and Advanced Planner
Agents
Autonomous: “Digital Workers” that perform tasks in real-time, and can then hand off to a human – Service Agent / Sales Dev Rep / Personal Shopper
Assistive: Augments humans by helping them perform tasks more efficiently – Agentforce / Sales Coach / Campaign Optimiser
How Does It Work?
There are three key components to Agentforce:
Data: Agentforce has advanced retrieval mechanics, allowing it to retrieve structured and unstructured data from multiple sources in real-time. Agentforce is able to review trends and preferences of each customer, quickly, whilst reducing the risks of hallucinations. As Agents become more autonomous, the importance of “good data” becomes more critical to the outcomes.
Processing: Agentforce is built on the Atlas Reasoning Engine and is a dynamic and proactive AI that can make deliberate decisions. It can react purposefully to customer questions and proactively take action to provide a seamless experience and follow up.
Action: Based on the guardrails that are set, Agentforce can then evaluate the data against the actions and tasks it has been set to then deliver responses that drive value, such as:
– Proactively contact leads and schedule meetings
– Respond to Cases with customised responses
– Manage marketing campaigns proactively without needing human input
– Build a Sales Plan for your team
What Else Do I Need To Know?
Salesforce takes its social and ethical responsibilities very seriously. There might be some hesitance when reading this article, and understandably so, what if an Agent goes rogue and does things that we don’t understand, or want it to do?
Salesforce Trust Layer ensures that all AI interactions are kept within your Salesforce organisation. The data you have, and the interactions you have with the AI never leave your system, and you can mask and delete prompts for further protection of sensitive information. Something to bear in mind if you use publicly available ChatGPTs for client research!
Each out-of-the-box Agent can also be configured with guardrails, to ensure that appropriate boundaries are put in place as to the data it can access, as well as the responses and actions it can take – you are in full control of the Agent, just as you would be when hiring a new staff member.
Next Steps
We hope you’ve learnt a bit about Agentforce during this article, and we’d love to see you at our roundtable that we’re hosting on the 13th November where we can answer any questions you might have and go into more details on Agentforce and AI in general. To register to join us CLICK HERE.
Written by:
James Cowley
Salesforce Consultant/Business Analyst
