Behind Salesforce’s 200 Prebuilt AI Agents for Workflows
Salesforce has quietly built up a library of more than 200 prebuilt AI agents for its Agentforce platform – which is a rapid expansion from the handful available when the company launched the service in October 2024.
The push comes as enterprises tackle how to actually implement AI agents.
Martin Kihn, SVP, Market Strategy, Marketing Cloud at Salesforce, says: “Customers and partners were excited and optimistic about AI and the agentic future, but there were challenges.
“One of the most common questions was: Can I really get this to work?”
Rather than dismiss these concerns, Salesforce’s product and engineering teams decided to tackle the problem head-on – by focusing on building prebuilt agents that could be implemented quickly and adapted without requiring extensive technical expertise.
Martin points out that: “Even the iPhone was mistrusted at first.”
Salesforce Agentforce showing early business impact
Salesforce’s strategy appears to be working – as since launching, Agentforce has delivered tangible results across different industries.
Engine, a workflow automation software provider, saw a 15% decrease in average case processing time.
Meanwhile, small business accounting firm 1-800Accountant managed 70% of administrative interactions autonomously during peak periods.
Brazilian media conglomerate Grupo Globo also achieved a 22% increase in customer retention.
Additionally, the deployment of AI agents has surged 233% in the past six months, according to the upcoming Slack Workforce Index.
More than 8,000 Salesforce clients have adopted Agentforce during this period.
Adam Evans, Executive Vice President & General Manager of Salesforce AI, says: “With Agentforce, we’ve unified agents, data, apps and metadata to create a digital labour platform, helping thousands of companies realise the promise of agentic AI today.”
Agentforce 3 addressing scaling challenges
Recognising that early adoption is only the beginning, Salesforce introduced Agentforce 3 in June 2025.
The update tackles the challenge of scaling AI agents across large organisations with new management and transparency features.
The Command Center provides complete agent observability – essentially allowing companies to monitor and understand how their AI agents are performing.
The update also supports the Model Context Protocol, a technical standard that allows different AI systems to share information and work together effectively, alongside more than 100 ready-to-use industry actions.
Adam says the company has: “listened deeply to our customers and continued our rapid pace of technology innovation.”
Martin adds: “Most of these prebuilt agents are intended to be a start, a relatively easy way for Salesforce customers to get over any initial reluctance and envision what a custom AI agent could do for them.”
The approach has unlocked creativity among users. “Once our customers became comfortable with the idea, there was no limit to the creativity they could apply and the business problems they could address,” Martin says.
Wrestling with AI agent identity
One of the thorniest issues enterprises face is determining how human-like their AI agents should appear.
Martin poses the question directly: “How human should agents be? There is some debate about where to draw the lines.”
Early chatbot development followed a simple playbook: make the bot seem human until it couldn’t handle the interaction, then seamlessly transfer to a real person.
“In the early days of chatbots, some builders assumed they would simply impersonate people until they could not, at which point a real person would pick up the thread – without the customer even noticing,” Martin says.
This approach has fundamental flaws. “We humans are good at detecting nonhumans; by definition, we are experts in humanity.”
He uses a customer service scenario to illustrate why this matters. “For example, imagine talking to a customer service agent about a health condition and they express some sympathy.
Thousands of Agentforce agents are already on the job, with many more to come
“You share some more. They’re patient and understanding, like Dr. Jennifer Melfi in The Sopranos. Then at some point, they say, ‘Thanks for sharing, I’m going to transfer you to a real person now.’ How would you feel?”
The solution, he argues, is transparency from the start.
AI agents should not masquerade as human employees or occupy traditional organisational roles.
“They should not be referred to as a ‘customer service associate’ or ‘rep’ but rather simply labeled by their work function: ‘customer service,’ ‘returns’ and so on,” Martin says.
“Today, thousands of Agentforce agents are already on the job, with many more to come. It’s easy to forget just how new this technology is.
“Our customers can sometimes feel like they’re learning how to drive an F1 car that’s already on the track.”
