logo
#

Latest news with #SandeepAlur

AI Agents Are Truly Here And We Need To Learn To Live With Them
AI Agents Are Truly Here And We Need To Learn To Live With Them

News18

time5 days ago

  • Business
  • News18

AI Agents Are Truly Here And We Need To Learn To Live With Them

AI agents are taking over our systems, and soon they will even enhance our jobs with more tech-enabled services. AI chatbots like ChatGPT started all the buzz in the market more than 2 years back, and now we are seeing the market move to the next best thing, Agentic AI or AI agents. You have used AI apps of all kinds, and even your regular apps now have some level AI-ness built into them. But AI agents are a completely new kettle of fish that is making its mark in positive and to some degree, worrying ways. Companies like OpenAI and Google are already making the big moves to get things right which is easier said than done when you have AI involved. Now that we have done a fair job of introducing AI agents, we'll let the experts do the talking and even telling you the nitty gritties in the world of these new agents. 'AI agents represent a new way of working — systems that understand intent, act across workflows, and adapt with context. se agents are built on advanced models, operate within clear boundaries, and are tailored for tasks from cybersecurity to human resources," explains Sandeep Alur, CTO, Microsoft Innovation Hub, India It is intriguing to hear Alur talk about tasks being handled by some of the regular verticals where a hands-on approach has been demanded for years. And it seems that more AI companies see these jobs becoming more handled by the tech than humans. 'Imagine now in large organizations there are 100 of people doing a certain task when the lead comes in, their job is to update the CRM etc. This can be now completely automated with Agentic AI," as highlighted by Dikshant Dave, CEO, Zigment, one of the companies to have deployed WhatsApp agents handling these tasks for its clients. In fact, a report by ServiceNow says agentic AI will catapult and reform the job market to generate over 10 million jobs in the next five years in India. AI Agents Everywhere These agents are programmed to operate tasks, and let humans manage the more complex part of their work or even regular stuff. For instance, you can ask the AI agents built into browsers to scan through a series of tabs and find you the best restaurant for dinner. 'It will be defined because it has a certain intellect, basis the conversation or whatever actions are happening. It takes a decision," Dave adds, giving more emphasis on its use and application across sectors. This definitely feels like the next phase of AI in our lives and the industry is gradually making it a core part of their structure. But does that come at the cost of human jobs? The boundaries are being already set, and the path is getting defined, which tells people where the AI threat to jobs will be less and where it will have a big impact. 'We can actually switch a mid-level manager or executive, a lot of the tasks that they were doing with an AI Agent. A lot of these underlying activities that were done, like the operation by the juniors are getting transferred to these agents," Dave opines. And the recent job cut trends align with his views, especially as you see the likes of Google and even Microsoft raise the pink slips for a variety of roles to embed AI further into their systems. Having said that, AI agents will only get some degree of control on the tasks, as businesses feel the pinch of adopting fully autonomous AI agents to handle their client work that earns them millions or even billions, as pointed out by Capgemini in its Rise of Agentic AI report. 'This is not about replacing people — it's about removing repetitive work and giving individuals more space to focus on creativity, problem-solving, and human connection," Alur shares his own insights. Even then, it clearly feels like companies are still getting the hang of the technology and would rather prefer a mix of humans and AI rather than fully let the tech do the bulk of the work. However, the trends will change quickly and by 2028 we might see more businesses get comfortable using AI agents and fully trust them to handle A to Z of their tasks. When it comes to business there is a very high degree of concern amongst customers about the data and the hallucinations that bring further trust factor issues in AI. 'We also kind of build a guardrails layer to ensure that there are zero hallucinations and the whole activity is that the LLM will be helm," Dave cites the best way to prevent such lapses. He also mentions that companies have to build another later on the top of the LLMs to ensure the accuracy of information. The risks clearly outweigh its benefits but he feels that with time we will have better checks in place to guard against such problems. 'Those concerns are happening and I would say that a lot of new ways and better ways will come up over the next few years where we will have a very stable system," he concludes our chat giving a positive outlook to the future even though AI will inflict some level of course correction in the larger scheme of things. tags : AI model view comments Location : Delhi, India, India First Published: August 05, 2025, 09:36 IST News tech AI Agents Are Truly Here And We Need To Learn To Live With Them Disclaimer: Comments reflect users' views, not News18's. Please keep discussions respectful and constructive. Abusive, defamatory, or illegal comments will be removed. News18 may disable any comment at its discretion. By posting, you agree to our Terms of Use and Privacy Policy.

APIs used to connect systems digitally, now AI agents will
APIs used to connect systems digitally, now AI agents will

Time of India

time18-06-2025

  • Business
  • Time of India

APIs used to connect systems digitally, now AI agents will

For decades, the humble API has been the plumbing that lets far-flung digital services speak to one another. A clean endpoint, a crisp schema, and the world's software snapped together like Lego. It's how your Whatsapp account connects to your Gmail service, and how most websites and applications talk to each other. Now that tidy requestresponse universe is being rattled by autonomous AI agents—programmes that do not merely answer a call but decide, reason and negotiate on our behalf. As the old choreography frays, software engineers now need to update their skills to help these agentic systems communicate effectively with each other. 'We're already seeing this shift building software at Katalon. Instead of rigid API contracts, we're designing how AI agents negotiate with each other—how they handle failures, share context, and maintain consistency across workflows,' says Coty Rosenblath, chief technology officer at the low-code testing specialist. 'Traditional APIs assume you control both endpoints, but with AI agents making independent decisions, protocols need to handle that uncertainty.' Microsoft's India innovation hub CTO Sandeep Alur agrees that APIs are ceding the stage. 'As AI agents become the new building blocks of software, the future of engineering is shifting from designing APIs to orchestrating intelligent agents,' he remarks, pointing to emerging standards such as Model Context Protocol (MCP) and Agent-to-Agent (A2A) communication that redefine how systems communicate, collaborate and scale. 'MCP provides a standardised way for AI models to interact with external tools, enabling real-time data access, tool integration, and reasoning across services. A2A enables structured collaboration between agents across platforms, clouds, and organisations, using shared goals, secure state management, and observable actions.' Essentially, A2A allows agents to talk to each other while MCP allows agents to talk to other software. Prompt engineering, once the darling skill of the ChatGPT boom, is already becoming passé. 'Prompt engineering was just the start,' observes Rosenblath. 'The challenge now is getting multiple AI agents to work together reliably.' Santosh Singh, president & global head of marketing & business excellence at Tata Technologies, puts it more lyrically: 'Prompt engineering is the spark—but agent orchestration is the symphony. The engineer of tomorrow will design conversations between intelligent collaborators.' From the vantage point of big tech companies, orchestrating an endless multitude of AI agents hinges on secure, multi-protocol integration, something software developers of all stripes should keep in mind during their next job interview. 'Future engineers must excel in designing multi-protocol integrations that are secure and adhere to boundaries or frameworks, preventing information leakage,' says Vishal Chahal, vice-president at IBM India Software Labs. He expects prompt templates, formulated by small language models, to direct agents towards task completion, shifting the engineer's craft from writing prompts to fine-tuning specialist models and weaving them into enterprise architecture. CHANGING SKILL SETS If the tools are changing, so too is the tool-box. Rosenblath believes success will belong to those who can break complex objectives into smaller tasks that agents can pursue independently but still align with an overall objective, designing how they share context, remember past interactions, and evaluate each other's outputs. Validation steps and feedback loops—long the preserve of quality assurance—become central to day-to-day engineering. IBM's Chahal unfurls a similar skills list: 'A solid understanding of how to automate tasks and processes, an understanding of machine-learning algorithms, the ability to craft effective prompts for AI models and APIs are crucial. Also, familiarity with integration tools is necessary for seamlessly combining AI capabilities with existing systems.' Vision, too, matters. 'Intuitive or imaginative skills to envision how the agent will respond in various scenarios is important.' Pavan Gurazada, adjunct lecturer at Northwestern University and senior faculty at Great Learning, divides tomorrow's workforce into tool builders and agent builders. The former will expose services as tools using standardised protocols like MCP. The latter will compose and orchestrate these tools to create agents that automate end-to-end workflows. Engineers, he predicts, will need to map complex business processes into agentbased workflows and ensure they align with evolving corporate goals. Sameer Goyal, head of engineering at Acuity Knowledge Partners, thinks systems thinking will be essential to understand how local interactions can produce emergent, sophisticated behaviors at scale. 'All things considered, organising agents is a confluence of distributed computing, cognitive science, and systems design rather than just an extension of current AI activities. Engineers who master this intersection will be highly sought after in the years to come.' And as much as this is still an evolving field, companies are already looking for these skills today. PaySprint's founder and CEO S Anand says they are actively on the lookout for these skill sets in potential hires for the company. 'We're looking for individuals that can function across evolving fronts — from AI logic and machine learning to business-aligned strategy and orchestration.' AI Masterclass for Students. Upskill Young Ones Today!– Join Now

DOWNLOAD THE APP

Get Started Now: Download the App

Ready to dive into a world of global content with local flavor? Download Daily8 app today from your preferred app store and start exploring.
app-storeplay-store