Forget multiple screens, keyboards, and mouse pads. Imagine a workday where you simply instruct your computer on the tasks you want done. Max Kirby, Principal AI Architect at Salesforce, believes this future is closer than you think. 

Kirby recently discussed his predictions for the agentic AI-powered workplace of 2030 with Marc Escobosa, VP of Salesforce Futures. He envisions a world with fewer screens, dedicated AI assistants, and instant access to the information you need.

Q. To talk about the future of agents, let’s start with their beginnings. When were agents first on the map? 

John McCarthy and Allen Newell discussed ‘intelligent agents’ as early as the 1950s. The International Journal of Agent Oriented Engineering has been publishing information on agents for the past two decades. 

But for many, it feels like agentic AI is moving quickly. That’s because technology is typically developed first, with practical applications — or what we do with it — figured out later.

What you’re feeling right now is the combined pull of decades of computer science thinking that’s already been done. All that groundwork means autonomous AI solutions like Salesforce’s Agentforce can be applied immediately to practical, real-world use cases — something that can feel like magic. 

All that groundwork means autonomous AI solutions like Salesforce’s Agentforce can be applied immediately to practical, real-world use cases — something that can feel like magic.

Max Kirby, Principal AI Architect at Salesforce

For example, with Agentforce, Precina Health is automating patient check-ins, unifying clinical data from EMRs and lab reports, and onboarding practitioners with coaching in rapport-building and clinical guidance‌ — ‌allowing clinicians to focus on high-impact care while scaling personalized diabetes support nationwide, saving thousands of dollars per year in reduced administrative overhead and clinical training costs. 

Q. Imagine it’s the year 2030. How do you think agents will alter our day-to-day work life?

Agents lessen the need for point-and-click interfaces. You give agents a task and they do it for you. Yet, for two and a half decades, UX design has focused on the human operation of the computer. Now, we have something that can essentially operate the computer for us, and that will change things.

The second thing that will change is screen time. A lot of our life is spent in front of screens, because we work at screens, are entertained by screens, and carry screens in our pockets. With agents, I predict we’ll spend less time working with screens and more time working with agent-friendly modalities, like Augmented Reality (AR) and Mixed Reality (MR), which naturally carry more context through speech, images, and video. 

Open Image Modal

Image Modal

Q. How will agents impact the jobs we’re doing?

A lot of today’s jobs don’t pass what I call the kindergarten test, which is, ‘Did your kindergarten self imagine that you would be doing this as a job?’ Take data entry — no one imagined in kindergarten that they’d be entering numbers into a computer all day for a career. 

Agents will change that — taking on the manual, administrative work so humans can focus on more creative, strategic parts of their job. In this way, AI will make us more human.

For example, we all know that CRMs are more effective with data added in. Yet, salespeople typically take three to seven days to enter data into their Salesforce instance after a customer interaction. When agents input that data instantaneously, without human help, reps can focus on more important work like relationship building or creatively solving a customer’s pain point with their products. Plus, the organization benefits from faster knowledge input. 

That kind of shift to strategic work is going to help all of us, and is why I’m a wild optimist about agentic AI. 

Q. By 2030, how will agentic AI change day-to-day interactions? 

By 2030, consumers will have personal AI agents acting on our behalf, handling everything from planning our schedules to handling our shopping and managing our finances. And businesses will use agents to talk to their agents, the proxies of the customer. Imagine, for example, a consumer’s personal assistant agent dialing into a retailer’s customer service center to initiate a return. 

This means that instead of designing personalization to cater directly to a consumer’s tastes, businesses will need to design personalization to align with the preferences and objectives of the consumer’s agent, or with the consumer’s agent’s interpretation of its human’s preferences and objectives.

How agents represent our personal interests will be drastically different, too. We send lawyers to go to law school to learn law. We send med students to medical school to learn to become a doctor. But there’s a limit to what they can store in their brains. Agents, on the other hand, have nearly unlimited space to store information. I’m not saying that agents will practice law or medicine in the exact same way as a human, and certainly won’t be as empathetic, but they’ll absolutely have that level of knowledge. Being able to pull that information out instantaneously will amplify us humans 10 to 100x.

Q. What’s an example of how an agent will better represent us in 2030?  

Let’s consider how agents will revolutionize the healthcare industry, specifically for nurses. Nurses are the backbone of our healthcare system, yet they are overworked and under immense pressure. In fact, one in four nurses is very likely to leave their current role. At the same time, nurses spend a significant portion of their time — between 12 and 25% — on documentation, primarily entering data into systems. This is time spent away from direct patient care, their core responsibility and where their expertise is most needed.

Imagine a future where agents automate this data entry for nurses on their behalf.

Max Kirby, Principal AI Architect at Salesforce

Imagine a future where agents automate this data entry for nurses on their behalf. Instead of manually inputting patient information, nurses can use voice commands or other natural language interactions to have agents record and update patient charts, medication schedules, and other essential documentation — approving and curating it instead of entering it manually. Beyond just data entry, agents will also be invaluable in providing nurses and healthcare professionals with instant access to critical information. Need a patient’s medical history, the latest research on a treatment, or information on drug interactions? An agent can retrieve it in seconds, allowing nurses to make informed decisions quickly and efficiently.

Additionally, we expect patient advocate agents to enter the market in the next five years. These agents will represent your interests as a patient, providing you with medical knowledge that normally requires years of schooling. Imagine having a friend who went to medical school by your side helping you through your care. Nice to have, but not everyone can access that kind of advantage. With personal agents acting as advocates, that will change.

This is more than automation of menial work — agents will level the playing field for us, making equalities in critical outcomes less impacted by asymmetries in knowledge.

Q. What are digital twins, and how will agentic AI change the way these work? 

A digital twin is a digital model of a real-life customer, product, or even business process. They allow us to replicate real-life scenarios to predict, test, and optimize strategies. 

Previously, building digital twins relied on structured data like fields and values, and required significant human involvement or complex software to set up and analyze the results. Now, with advancements in agentic AI, and systems that can blend structured and unstructured data together at petabyte scale, like Salesforce’s Data Cloud, agents can interpret the semantic meaning within unstructured data — like customer conversations — to add rich context to digital twins. This allows businesses to understand customers in their own words and predict their needs.  

Open Image Modal

Image Modal

Bumble recently proposed using digital twins to predict two people’s compatibility. This same concept, backed by unified data and agentic AI, can be applied to business scenarios like testing ad campaigns or pricing strategies. For example, a marketing team could use a digital twin of their target customer to simulate how they would respond to different ad variations, predicting which one would yield the highest engagement before launching the campaign. 

Ultimately, digital twins, enhanced by AI’s ability to process diverse data types, offer a powerful tool for enhancing understanding and decision-making. I expect we’ll see more of this in the coming years as businesses work to predict what their customers need.  

Go deeper: 

• Explore content from the Salesforce Futures team 
• Learn why conversations and context unlock agentic AI impact.
• Learn how to prepare your organization for agentic AI
• Read why metadata is key to building an AI agent ecosystem