From Chatbots To Taskbots: What is Agentic AI And How Can It Help Businesses?
NDTV Profit breaks down what 'Agentic AI' means and how it can be applied in organisations to streamline convenience and productivity.
Agentic AI is one of the latest artificial intelligence innovations to have a lot of buzz in the industry, with companies from Infosys to Google racing to design and market agentic AI offerings to governments and businesses.
NDTV Profit breaks down what 'Agentic AI' means and how it can be applied in organisations to streamline convenience and productivity.
Agentic AI are task-orientated artificial intelligence systems that are geared towards planning and executing specific goals. Unlike their generative variants that are aimed at creating content from user inputs, agentic AI formulates solutions in real time and executes them.
How Agentic AI Works
Agentic AI uses five major automation patterns to undertake operations, according to Arenas.
The first is the 'Tool-Use Pattern', which involves the AI directly interacting with the systems that the organisation uses to undertake data retrieval, workflow creation, interfacing with other software applications, and executing transactions.
The second is the 'Reflection Pattern', where the AI goes back and reviews its output and catches errors to improve upon without human input, which can help it eliminate errors before they are even made.
The third is 'Planning Pattern', where the AI uses the previous two patterns together to break down complex processes into tasks that it can take action on.
"Planning agents address this (complexity) by breaking high-level goals into actionable tasks, tracking progress, and adapting as requirements shift," Arenas said.
This brings down costs for processes while automating a majority of them.
The fourth is 'Multi-Agent Pattern', where, as the name suggests, the AI agents communicate with and collaborate with various other agents with specialised tasks to execute processes with a lot more speed.
The fifth and final pattern is the 'ReAct' or reasoning+action element, which has the AI resolve roadblocks in real time instead of depending on pre-set scripts.
"ReAct agents alternate between reasoning and action—taking a step, observing results, and deciding what to do next. This allows agents to adapt to ambiguity, evolving requirements, and situations where the best path forward isn’t clear," Arenas said.
What It Does
According to Yina Arenas, Vice President of Product, Core AI of Microsoft, agentic AI uses 'retrieval-augmented generation' technology, which provides answers to questions and presents detailed information requirements at unmatched speed.
This technology has applications in business operations just as it does with AI chatbots, helping to iron out major wrinkles in the information classification and organisation.
This includes addressing the execution flaws in form submissions, record maintenance and more complex actions executed across many different systems, which are mainstays of most enterprise processes.
Agentic AI addresses these issues via planning, adapting to novel situations and coordinating across systems in real time, which traditional automation systems are unable to do.
"Traditional automation tools—scripts, Robotic Process Automation (RPA) bots, manual handoffs—often struggle with change and scale, leaving teams frustrated by gaps and inefficiencies," Arenas stated in her blog post.
Applications
These processes are chained together, having agentic AIs combine them to execute tasks, keep records and troubleshoot in real time, making them ideal for business and governance.
