Cognigy: What is Agentic AI and Why Does it Matter to You?

Niall Carter, Assistant Vice President (AVP) UK&I at Cognigy, says: “everyone is talking about agentic AI,” the problem is, not everyone knows what it is and how to utilise it.

Gartner has predicted that by 2028, 33% of enterprise software applications will include agentic AI.

Analysts at IDC report that 70% of APAC organisations expect agentic AI to disrupt business models within the next 18 months.

Meanwhile, financial experts have forecast that the global agentic AI market is to reach US$196.6bn by 2034. 

But what is it? What does agentic AI mean or do? And how is it different from the other forms of AI?

Niall speaks to AI Magazine to fill in the gaps about this emerging AI and its possibilities for business practices.

The next generation of assistance

For the last few years, it seems that the world has been focused on Gen AI and its capabilities – but Niall believes that there has been a core limiting factor.

“Gen AI is fundamentally a passive technology,” he explains.

“It requires prompting. It will create the most human-like exchange or breathtaking art, analyse your data or write your code, but only in response to an instruction.

“There’s logic and comprehension under the hood. But the actual execution is a reactive transaction, even with recent iterative and multi-modal capabilities.

“This means that autonomy is what sets agentic AI apart,” he says, “as agentic AI describes the use of AI-powered agents and digital assistants that can observe, reason, plan and act without human prompting.

“Rather than responding to specific instructions, they are goal-oriented.

“This means they pursue a goal proactively until it is achieved, breaking it down into logical steps, just like humans do.

“They can do so because they are integrated with back-end systems, tools and knowledge and given executional permissions, such as triggering workflows, retrieving and inputting information, or sending messages.

“This all adds up to an agent that can actively fulfil a role, not just singular functions.”

A giant leap forward

Niall explains that agentic AI could be perceived like workflow automation applications that use AI, yet those are examples of deterministic systems.

“If something is deterministic – AI or otherwise – it gives you consistent outputs or results because it follows pre-determined rules and logic,” he says.

“It’s largely what programming pre-AI has been. It’s reliable.”

Yet, there is a “defined, or rather a determined, path of instruction when it comes to AI,” Niall says – “and it’s what makes deterministic AI more explainable and transparent, which is useful for precise or science-based use cases.”

He adds: “But most real-world tasks aren’t this linear, this neat.

“Most tasks are more conceptual and goal-oriented.”

Niall believes that for digital assistants to have real value, they need to be able to interrogate and break down a problem, factor and plan, juggle multiple tasks, combine and build upon them – even to collaborate with other AIs.

“Most importantly, they need to be self-starting, autonomous and proactive,” he says.

“This is the difference between deterministic systems and agentic AI.

“AI agents may even use deterministic tools to achieve their goal, but the difference is that agentic AI can decide when, how and why to use them — operating autonomously, adapting to changing circumstances, pursuing and reviewing outcomes without needing to be prompted at every step.”

Agentic in action

Niall observes that customer service and support teams have been some of the earliest adopters of agentic AI.

“By upgrading old chat-bots that only respond to FAQs and rely on keywords, AI agents use Gen AI and combine it with the ability to follow structured processes to provide human-like, goal-oriented interactions,” he says.

AI agents in customer service can:

• Provide service updates or issue refunds
• Book flights or insurance
• Chase up invoices
• Account for customer preferences, external variables and company policies

“They also have persistent short or long-term memory (another distinguishing feature from traditional Gen AI),” Niall explains, “enabling them to continue conversations and tasks over multiple sessions and across channels.

“These characteristics make agentic AI an incredibly versatile and effective tool – one that really blossoms at scale. Its actions translate to measurable, value-based metrics and demonstrably relieve human teams of repetitive, high-volume tasks.”

Further solidifying agentic AI’s impact on customer service, Gartner believes that agentic AI will autonomously resolve 80% of common customer service issues without human intervention by 2029.

Niall points out that Bosch has already integrated more than 90 agents across its business units, spanning customer service through to HR.

The company describes its agents, created using Cognigy’s platform, as “freeing up the time for employees to think and work more strategically, taking over a lot of administrative processes and mak[ing] their lives easier with agents”.

“Agentic systems embody a new kind of human-AI collaboration,” Niall concludes.

“They’re making companies more resilient, agile and productive – enabling staff to focus on more satisfying, value-driving activities – and providing customers with satisfying and responsive experiences with no wait times.

“Agentic AI is a big deal.”