Technological advancements are known for revolutionizing the way humans work.
Steam power ignited the first Industrial Revolution, moving workers from fields into factories. The second Industrial Revolution, powered by electricity, ushered in assembly lines and mass production. The third was characterized by electronics and computers, and laid the groundwork for the digital age. Now, we find ourselves in the Fourth Industrial Revolution, which is defined by interconnected technologies like big data, robotics, and AI.
While AI research has been underway for nearly 70 years, its big break has finally arrived. ChatGPT turned what seemed like science fiction into an everyday reality. Generative AI is now pervasive, diagnosing medical conditions, optimizing supply chains, qualifying sales leads, and more. This breakthrough also set the stage for agentic AI, a technology that can not only generate content but make decisions and take action with limited or no human supervision.
Through natural language processing (NLP) and powerful reasoning engines, autonomous AI agents mimic human behavior, making them ideally suited to handle a wide range of intricate and ever-changing situations. Agentforce, the agentic layer of the Salesforce Platform, provides a mechanism for quickly developing and deploying customizable agents to handle a broad range of business use cases, from order status and tracking to scheduling and appointment management.
How agentic AI unlocks a digital labor force for businesses
For businesses, this shift has profound implications: The possibility of a digital labor force working alongside humans to streamline operations, enhance productivity, reduce costs, and unlock new levels of innovation and scalability. For the first time, workforces can exceed the bounds of human capability with trusted, autonomous AI agents working 24/7 to augment and greatly expand productivity, efficiency, innovation, and business competition.
“Agentic AI is a new labor model, new productivity model, and a new economic model,” said Salesforce Chair and CEO Marc Benioff.
Agentic AI is a new labor model, new productivity model, and a new economic model.
Marc Benioff, Chair, CEO, & Co-Founder, Salesforce
With labor markets the tightest they’ve been in two decades, new labor models are sorely needed. Indeed, an estimated 85 million jobs could go unfulfilled by 2030 as populations age, birth rates decline, and worker expectations change.
With digital labor, companies increase output and productivity without increasing headcount. Take small and medium-sized businesses that typically don’t have the resources to scale and compete with larger enterprises. With agents, these businesses can add digital workers that require little to no supervision, taking on activities like lead management and qualification that fuel growth without adding major operational costs.
This transition away from human-dependent workflows creates an opportunity for 24/7 productivity, enabling businesses to serve global markets without local teams or infrastructure.
The result could be a fairly significant uptick in global GDP, Benioff said. In fact, Goldman Sachs predicts a 7% increase in global GDP in the next 10 years because of AI.
How digital labor will change work
Each industrial revolution introduced new, more specialized roles and teams — like machine operators, technicians, and supervisors — while increasing productivity and growth. The addition of autonomous AI agents to workforces promises to fundamentally alter operational and organizational structures, creating hybrid workforces of human and digital labor and entirely new ways of working.
“Digital labor is a new horizon for business … How we architect our businesses and run our businesses and staff our businesses and think about our businesses will never be the same,” said Benioff.
Digital labor is a new horizon for business.
Marc Benioff, Chair, CEO, & Co-Founder, Salesforce
Operationally, companies will change how work gets done, offloading complex repetitive tasks to agents so human employees can focus on higher-priority matters. For example, for most companies, service reps have to answer frequently asked questions, process refunds, and troubleshoot technology issues. AI agents can take on the majority of these routine tasks and workflows, changing the role of a human service rep to handle only the most complex cases, or supporting teams training and refining the AI models to improve accuracy and responses. This will shift human work from execution to creative strategy and oversight.
AI agents aren’t just tools people will use occasionally; they will be integral collaborators in this new way of working. For example, an AI agent might handle an initial customer service inquiry, but upon sensing frustration in the customer, flag it to a human agent for a personalized response or intervention. Afterward, the service representative would provide feedback on how the AI agent handled the case, helping the AI improve over time and reducing the need for human intervention on similar cases.
As a result, AI agents will change organizational structures, evolving existing roles and teams and introducing new roles in areas like AI agent management, AI risk and governance, AI operations management, AI training and development, and AI workforce integration.
“We’re in the biggest workforce transformation of our lifetime as we unlock the power of agents and humans working together,” said Salesforce Chief People Officer Nathalie Scardino. “Every organization will be called to redesign their people strategies, redeploy talent to support a future workforce with agents, and reskill employees — and every employee will need to lean in on human, business, and agent skills to drive success for themselves and for their customers.”
Scardino continued, “It’s a new generation of skills. As with other technological revolutions, AI will create new jobs and opportunities – and already has. We’re hiring for jobs today that didn’t exist as recently as last year – like ethical AI architect or agent product manager – and we’re reskilling employees through tools like Career Connect so they can jump into roles that are helping shape our future with agentic AI.”
Embedding agent trust across the organization
Trusting the technology is key to integrating agents. That’s important because 93% of global desk workers don’t consider AI outputs completely trustworthy for work-related tasks.
“You wouldn’t hire an employee you don’t trust,” said Rob Katz, who co-leads a team responsible for operationalizing responsible AI and ethical technology at Salesforce. To Katz, trust comes down to three things. Employees need to see what the agent did (transparency), why the agent did it (explainability), and know what to do next (control).
You wouldn’t hire an employee you don’t trust.
Rob Katz, VP, Product Management – Responsible AI & Tech, Salesforce
Sixty percent of consumers say advances in AI make trust even more important. Companies should implement rigorous testing methodologies to ensure AI systems operate fairly and equitably across various cultural and social contexts. Salesforce’s Trust Testing, for example, systematically uncovers and mitigates subtle biases by incorporating diverse user perspectives.
Building a technical foundation for digital labor
Of course, every AI transformation starts with preparing the underlying technology. Organizations need a system for connecting valuable business data and metadata to give agents the content and context they need to be effective.
“When we keep focus on the data quality, [the agents] get better and better,” said Prakash Kota, CIO of Autodesk, a Salesforce customer using AI to help employees focus on the highest value work.
This should include a hyperscale data engine like Salesforce Data Cloud, which collects and harmonizes all of the data and metadata agents need to produce insights and take actions grounded in customer data. With Data Cloud, AI agents not only have access to every relevant piece of trusted enterprise data — both structured and unstructured — but also understand its context, enabling agents to deliver meaningful recommendations and actions in real time. Through Data Cloud’s Zero Copy Partner Network, companies can connect their existing data lakes and warehouses to ensure all relevant business data feeds into the agent.
The agent is nothing if it hasn’t got relevant content it can pull from.
Bernard Slowey, VP, Customer Success, Salesforce
“The agent is nothing if it hasn’t got relevant content it can pull from,” said Bernard Slowey, VP of Digital Customer Success, who oversaw the deployment of Agentforce on help.salesforce.com. Since implementing, Salesforce is handling about 32,000 customer conversations per week with a resolution rate of 83%.
A unified platform enables users to automate a broad variety of tasks and workflows across departments. With Salesforce’s Agentforce, for example, a user will interact with one interface that spans all Salesforce applications, including Slack and Tableau, and can generate customer campaigns, handle customer service cases, answer questions, and create and summarize content, all in the flow of work. This will be important as more and more agents enter the workforce, requiring collaboration among agents, not just with humans.
In the agentic world, orchestrations of agents will tackle more complex challenges, like building sales or marketing campaigns, that typically require the involvement of multiple business disciplines. Unlike simple copilots, multi-agent systems will collaborate with one another, adapt, and execute in concert.
Salesforce Chief Scientist Silvio Savarese offers an example: Imagine a customer’s personal AI agent negotiating with a rental company’s AI agent. The customer’s agent optimizes for the best price and value, while the rental company’s agent aims to maximize revenue through add-on services. The business agent must balance aggressive sales tactics against the risk of losing the deal to competitors. Like APIs that connect different software formats, a system will be needed for enabling such multi-agent interaction. A platform or system-level architecture like Agentforce can handle any volume of requests from agents representing human interests.
Being able to test, monitor, and report on agent activities is another important element. Just as human supervisors conduct regular performance reviews, digital workers will need to be tracked and tested to make sure they’re operating at their best. Salesforce’s Agentforce Testing Center enables organizations to easily test AI agents using synthetically generated data, ensuring accurate responses and actions, with complete monitoring of usage and feedback.
Once deployed, Salesforce experts recommend using a scorecard to monitor and report on agent activities. For example, Slowey’s team deployed an agentic scorecard with key KPIs like agent conversations completed, response time, and escalations deflected.
Looking ahead
As agentic AI continues to evolve and reshape industries, the path forward is clear. Businesses must reimagine their operational and organizational structures and adopt agentic technologies to harness the full potential of a digital labor force. This will mark a threshold moment in the Fourth Industrial Revolution, where seamless collaboration between human and digital workers becomes the new normal.
The most forward-thinking organizations won’t just adapt — they’ll lead, shaping a future where human ingenuity and digital intelligence work in concert to drive unprecedented efficiency, creativity, and growth. Those who embrace this shift now will define the next era of work.
Go deeper:
- Learn more about Agentforce and how Salesforce is using its own agent technology to power help.salesforce.com
- Read why bad data is like junk food for AI
- Learn 10 collaborative skills needed for teams to succeed with AI agents