By Elizabeth Anderson
With the benefits of traditional AI now clear, the next wave of innovation is coming from agentic AI – where various AI tools can work together, acting autonomously, to make decisions and solve problems.
Agentic AI goes beyond benefits such as automating more complex tasks. It can suggest improvements by collaborating independently with other parties. It is also able to make decisions about when human judgement is required when something doesn't follow the standard path.
Agentic AI is already in use in many applications but the full potential will become clearer over the coming years, explains Bryan Zhang, executive director at the Cambridge Centre for Alternative Finance. He says current AI still often relies on explicit human prompts, passively responding to requests. Agentic AI changes that equation.
“Agentic AI not only generates content but also perceives, learns and takes action via integration with tools with minimal human involvement based on set goals, enabling continuous adaptation and decision-making. Agentic AI orchestrates multiple agents using large language models as a collective brain to solve complex, multi-step problems autonomously.”
Saving on admin and winning new leads
Gordon Reid, founder of digital studio Middle Boop, is one entrepreneur seeing the benefits of agentic AI. He says the technology saves him hours in admin and is winning him new business.
“We're using tools to automate a lot of our cold outreach, network building and to help us discover potential leads that we feel we can offer genuine value to. It's brilliant,” he says. “The tools run all the time and take the pain away from me having to do it myself. The framework we‘ve built is hooked up to my calendar so if any meetings are booked or rescheduled, I just need to prompt it, rather than manually arranging everything. We also use it day to day for sending relatable instant replies for any enquiries we receive.”
Reid adds he’s also using agentic AI to adapt content. Previously, content would start as an article then painstakingly adapted for Linkedin posts, other social media, blog posts or emails. “Now we use one tool which adapts it to all of the different channels, plans and sets up the posting in a content calendar and takes all of the admin out of it, leaving us more time to think of more ideas for content creation,” he adds.
Feeling inspired? Here are some other ways agentic AI could be deployed:
Spotting issues such as an unexpected drop in revenue due to a payment delay from a key client – and recommending the appropriate action.
Aaron Harris, Chief Technology Officer at Sage, says customers are primarily using its AI assistant Sage Copilot, which is evolving to incorporate agentic AI capabilities, to automate the routine, repetitive workflows in accounting such as receiving and paying invoices.
For example, Haneker, a London-based specialist in premium architectural joinery, is using Sage Copilot to deliver insights into income, spending and cash flow, and receive automated alerts and surfacing early warnings on project overspend. This lets Haneker make smarter, faster decisions without getting lost in spreadsheets or delayed reports.
“Sage Copilot flagged a spike in supplier costs, saving us time and money by catching an ordering error early,” says Margaret Sadzynska, Managing Director at Haneker.
Agentic applications are currently being developed to take Sage’s AI capabilities even further. Sage’s next generation of AI agents will be able to orchestrate complex workflows across multiple systems.
Within the next 12 to 36 months, customers will be able to access agents with true autonomy: systems that can detect business risks, initiate multi-step responses, and communicate with other agents to resolve issues proactively. As an example, customers could ask questions such as ‘What caused the drop in Q4 revenue?’ and receive detailed answers based on AI analysis of multiple systems. This will help them make more informed business decisions to support growth.
However, not everything is simply handed over to machine learning. AI also knows when to bring in a human if there’s an anomaly or an important decision, and this will continue as agentic AI is rolled out.
Humans still have an essential role to play, says Harris, and he is very bullish on the accounting profession becoming more valuable in the future. “AI is not going to sign an audit report. We’ll always need humans to provide assurance. Most of the businesses we work with, when making efficiency gains by using AI, direct staff away from repetitive admin to more valuable tasks to take on more strategically important activities.”
What’s holding businesses back?
Complexity remains a barrier to many businesses looking to take the next steps in utilising the full potential of AI. Only 1% of enterprises view their generative AI strategies as mature, according to McKinsey’s Seizing the agentic AI advantage report.
Adam Williams, of North Shields-based artisan cheese manufacturer Tyne Chease, says he uses AI to save time on administrative tasks but does not know enough about how AI can be used fully. “We’re not knowingly using agentic AI but it’s something I’d love to incorporate into this business. I can definitely see the benefits,” he adds.
To get the most from agentic AI, where issues can be detected and coordinating action is taken autonomously, key business data such as invoices must be correctly structured to be machine readable. This is where small and mid-sized businesses (SMBs) face barriers currently and why agentic AI is still in relative infancy when it comes to practical use cases.
There is also a lack of skilled workers able to keep pace with the fast-changing technology. Almost half of small businesses say they or their workforce lack the knowledge or skills to utilise AI effectively, according to a recent report from the Federation of Small Businesses (FSB).
Yet AI tools are making it easier for businesses to get access to sophisticated technology that was previously only available to large, well-funded organisations, says Harris. Even a micro business with one employee can benefit from advanced AI tools.
“I would say that AI is the great democratiser of technology because it’s accessible to the smallest organisations, not just the big players. Our job is to deploy AI in such a way that you don't need to hire data scientists or AI engineers. It means you don’t need AI expertise in your company to still benefit from what it offers,” he adds.
