A woman records a voice mail on a smart phone using using AI application.
Imagine a bunch of nerds in the 1960s, sitting around thinking: “Hey, what if we could send messages between these things called computers instead of just passing notes in class?” They proceed to further think: “Let’s make these machines talk to each other so we don’t have to!" And boom, the internet was born—a product of pure laziness, fuelled by an overabundance of caffeine and an allergy to exercise.
So, in 1969, these geniuses at the US Department of Defense created ARPANET (Advanced Research Projects Agency Network), which was basically the internet’s awkward teenage phase. Picture it: slow, clunky computers passing messages like, “Hi there, got your file, LOL,” over cables that would make today’s Wi-Fi laugh out loud. But, hey, it worked!
The internet grew up, put on a nice suit, and was ready for the world. It became more like a crowded party where everyone’s invited—websites, email, chat rooms! That’s when the World Wide Web showed up like the cool kid, making everything more accessible. Now people could share cat videos, awkward family photos, and conspiracy theories, all with the click of a button. What a time to be alive.
Game changer
And then came artificial intelligence or AI. Essentially the internet on steroids, it now runs the show now. It can read your emails and manage your schedule. It can also analyse your business data, handle customer service, and basically run your whole company while you’re having a barbecue in your backyard.
“It’s not about the tech itself; it’s about how you use it to change the game for everyone,” Adam Evans, an AI guru and Senior Vice President of AI Platform Cloud at Salesforce, a cloud computing company, told an online conference hosted on September 18.
Flashback to 2016, when AI first hit the scene with its crystal-ball powers. It was like a supercharged fortune teller—able to predict future events just by crunching past data. Whether it was forecasting stock prices, customer churn, or even tomorrow’s weather, executives were obsessed. AI became their go-to for figuring out what the future held.
Fast forward to 2023, and AI didn’t just stop at predictions; it started creating. Not just any stuff, but things like new chemical compounds, futuristic video game worlds, and even logos. It learned from real-world data, then used that knowledge to conjure new content out of thin air—whether it was music, 3D models, or hyper-realistic images for virtual reality.
It was a dream come true for the curious minds. AI was no longer just a helper—it was an innovator, cooking up things that didn’t exist yet, pushing the boundaries of what tech could do.
"Ten years ago, it took years and large teams of engineers to build systems that could analyse emails or suggest next steps. Today, thanks to generative technology, we can achieve the same results in just days with fewer resources,” says Mr Adams. “It’s an entire collapse of companies and complex products into streamlined features.”
This era has witnessed companies thriving through increased profit margins and heightened productivity, according to findings from respected business audit firms like PricewaterhouseCoopers and Deloitte. The challenge now is how to stay ahead when there’s always more work and never enough time.
AI agents
Mr Adams explains that the solution lies in AI agents—custom-built, tailored to your needs, and constantly learning from your data. These agents allow businesses to get more done, save time, and accelerate growth. But who, you ask, is this magical agent? At its core, it’s a reasoning engine—a brainy blend of context and action that makes decisions on the fly. Think of context as your secret sauce: it’s your customer info, your policies, and your instructions, all connected to this engine. If it were just data, sure, it could answer questions, but we’re not stopping there. We want action!
AI agents are computer programmes designed to perform specific tasks or functions by mimicking human-like reasoning and decision-making. Combine your data with actions built on the platform and sprinkle in that reasoning engine, and voilà! You’ve got an agent ready to unleash a world of possibilities.
Imagine an artificial sales agent that not only follows up with leads and qualifies them but also hands them off to your human sellers like a relay race pro. There are also service agents—your round-the-clock warriors, tackling customer inquiries and deflecting tier-one tickets faster than you can say “customer satisfaction.” They can transcribe conversations, tag customer profiles for marketing campaigns, identify product defects for your supply chain, and ensure compliance with business policies.
Picture this: you’ve placed an order online, and now you want to get in touch about a little hiccup. The data here is like your customer profile, your order details, and the company policies all wrapped up in a neat package. The actions? Well, they’re the nifty tricks the agent uses to pull up your profile by email or phone, or to check that order status using the order number.
So, let’s say a customer chimes in with, ‘I haven’t seen my order yet my order number is X.’ What happens next? The agent swoops in, grabs that order number, and dives into the data. Depending on what it finds, the next step unfolds. Maybe it checks up for it in the inventory. Is it out of stock? If it is, the agent pivots and refers to the company policies on how to handle such a situation.
Now, while that’s a simple example, the real world is far more complex. That same line—‘I haven’t seen my order yet; my order number is X’—could trigger countless different scenarios.
Mr Adams, a programming whiz with 30 years under his belt, illustrates this perfectly. “It’s about connecting actions and data with logical reasoning, step by step, reacting to the information as it rolls in to navigate the choices.”
The new programming
Ah, programmers of yore had it rough! They would spend countless hours constructing elaborate flowcharts, mapping out every possible conversation scenario. Imagine a maze of code—practically impossible to navigate! But now, with AI's reasoning engine, all those tedious paths are traversed in real-time.
What however happens when the agent hits a snag? Enter the humans! Rather than being replaced, they join forces with AI, tackling the trickier customer queries that the bots can’t manage.
“Agent force is what AI was meant to be,” proclaims Dr Gary Brandeleer, a commercial engineer. “It’s AI working for your company, customers, and employees. It takes on the repetitive tasks, letting your team focus on what really matters—high-impact work and exceptional customer experiences. Plus, it unleashes creativity like never before!”
Navigating AI in enterprises has often felt like chasing a mirage: higher productivity, fatter margins, and lightning-fast answers from AI-driven customer success. But many programmers warn that this is just scratching the surface of what’s really going on beneath the waterline.
“While everyone’s swimming happily, a few sharks lurk below,” Dr Brandeleer cautions. Companies are grappling with hidden costs, disjointed data, and ever-evolving models. The kicker? The industry keeps insisting you build your own AI, but by the time you’re done, it’s likely already outdated—like trying to catch up with the latest fashion trend that’s just gone out of style.
Trust, business value
So, how do we let these AI agents run the show while still delivering real business value? Avanthika Ramesh, Director of Product Management and a software engineer at Salesforce, has the answer: your AI is only as good as your data. That means your agents need access to trusted, accurate, and real-time information.
“Just a year ago, we thought the magic bullet was to train a custom model on all our business data,” she explains. “But soon we realised that was a time-consuming, costly endeavour. Those models can become outdated faster than a trendy haircut, especially as your data evolves.”
At its core, these AI agents operate by firing off a series of prompts to a large language model that generates responses.
“It’s all about delivering the right data to the right prompt at the right time,” Ramesh says.
But not just any data will do; it has to be structured correctly. Luckily, both structured and unstructured data can coexist in one system, allowing your agents to efficiently search for and retrieve what they need when the moment strikes. Erin DeCesare, Chief Technology Officer of ezCarter—a company that connects businesses with restaurants and caterers through an online marketplace—offers insight into this. Her team has been harnessing AI and machine learning to mine their 17 years of catering data.
“We’re tackling what we call the discovery phase,” she explains. “Today, the solution hinges on customers picking their favourite restaurant or chatting with our super helpful human customer service agents who have the entire catering ontology mentally catalogued.”
So, when a customer says, ‘Can you order me something like what I had last month?’ these agents can decode that unstructured prompt and understand what the customer craves.
“As we scale up, we need to enhance our agents with a solution that provides that expert feel,” DeCesare adds.
Actions
Chair Cheng, Vice President of AI Science & Engineering at Salesforce says that a suite of autonomous agents designed to tackle common tasks for service teams, sales, and marketing can be easily customised to fit your business needs, driving success like a well-oiled machine.
At the heart of this technology is the Atlas Reasoning Engine, which mimics human thought processes. It evaluates user requests, refines them for clarity, and fetches relevant data from your business, CRM, and data cloud to tailor plans for specific challenges. This dynamic process continually refines actions to ensure relevance.
“With this powerful engine, if you can describe a task, you can get it done. Customise out-of-the-box agents or build your own using Agent Builder,” Cheng explains.
How do you remain relevant?
At the State of the Digital Economy Conference in Kampala this August, Dr Liz Ngozi, founder and CEO of The International Social Impact Institute and Adjunct Assistant Professor at New York University, emphasised that solo entrepreneurs, small businesses, and under-resourced nonprofits can leverage AI to craft compelling proposals, enhance resumes, and customise various content like surveys or reports. She views these tools as efficiency boosters that save the most precious resource: time.
“If you’re a salesperson, your time should be spent selling, not generating reports. That’s the exciting potential of this technology,” she noted. “The best results come from clear prompts […] We must embrace these tools. They are essential for progress.”
