From Commands To Conversations: Why Agentic AI Feels More Like A Partner Than A Tool
Alan Turing led us on a path where machines became force multipliers to human endeavour. And here we are in the world of Agentic AI.
By Kumar Somya Pandey
It's November 1940, and Hitler’s Luftwaffe, the German air force, is conducting ferocious night raids on London. The fate of civilisation hangs in the balance. France has already surrendered, and Britain is the next domino to fall. The British have not been able to decipher the code that the Germans are using to time their raids.
Alan Turing, in Bletchley Park, 55 miles from London, is creating ‘Bombe’, a code-breaking machine that would not only greatly contribute to the Allies' winning the war against Hitler but also plant the seeds of the AI revolution that we find ourselves in today.
The German ‘Enigma’ cipher - a code that changed daily with over 150 quintillion combinations! The ‘Bombe’ ran millions of permutations and combinations, providing cryptographers with the brute force needed to narrow down combinations of the code.
The Bombe wasn’t learning; It still needed human intervention. It was still just a tool.
The advent of Agentic AI, aided by epoch-making progress in deep learning and computing power, renders enterprise software in its present form passé.
Agentic AI brings speed and accuracy to decision-making. Businesses that refuse to get on board and treat Agentic AI as just a tool, not a partner, would be left in the lurch.
From Tool to Partner: The Rise of Agentic AI
C.A.C: The three pillars of Agentic AI
C = Context - Agents are always context-aware. They have memory.
A = Autonomous - Based on system permissions, agents are autonomous to make decisions.
C = Conversation - the interface to interact with agents is based on natural language. Agents learn your language.
From Principle to Practice
Now, how does all this translate into real-world applications? As the bard said, 'an ounce of practice is worth a pound of theory’.
That’s what we are aiming to achieve here: How deployment of Agentic AI in the workflows of companies is bringing speed and accuracy to decisions.
A hospitality major running 125 hotels in India, by the sheer size of its operations, dealt with thousands of invoices daily. The vendors would send invoices via WhatsApp and email, and not always accompanied by Purchase orders (PO). Mismatches between GRN (Goods Receipt Note), PO, and invoice were routine. As a result, vendors’ payment delay became a commonplace, yet unpleasant, feature of business.
At this critical juncture, the company decided to deploy Agentic AI into its finance workflow - the ‘Payables agent’ and ‘Ask Me Anything’ (AMA) agent. Looking from the prism of C.A.C., this is how agentic AI was deployed:
Context: The Payables agent has contextual memory. When a vendor sends in an invoice, the agent cross-references it with a matching PO. Further, it analyses the vendor’s past pricing trends and delivery timeliness, and if there is a sudden spike, it flags the same.
Autonomous: When an invoice comes in, it is validated in real time with GRNs and PO’s and payments. If there is something amiss, like a mismatched quantity with GRN, then the finance department is notified to take action. An upshot: invoice processing time fell by 47%.
Conversation: The finance team uses the AMA agent to generate highly contextual information in a natural language format. Queries like: which linen vendors had the maximum pricing variance over the last 2 quarters?
The Payables and AMA agents are just two agents that have been categorically deployed. There is a full stack of agents ready to be deployed across functions for the hospitality major.
The Brave New World of Agentic AI
Alan Turing led us on a path where machines became force multipliers to human endeavour. And here we are in the world of Agentic AI, no longer force multipliers but agents of productivity revolution.
In a biting critique of culture, the American poet Gil Scott-Heron declared, ‘Revolution will not be televised.’ As it stands, revolution is agentic AI, and it will be productised.
(The author is the Chief Strategy Officer at HostBooks Limited)
Disclaimer: The opinions, beliefs, and views expressed by the various authors and forum participants on this website are personal and do not reflect the opinions, beliefs, and views of ABP Network Pvt. Ltd.
Top Headlines
