ব্যান্ডউইথের বাইরে: এজেন্ট-ভিত্তিক এআই-এর জন্য ইন্টারনেট পুনরায় কনফিগার করা
The internet is moving towards a fundamental shift that will change the way we connect and design networks. At the heart of this transformation is the rise of agent-based AI: autonomous systems that operate, make decisions, and transact on the internet—not as proxies for human clicks, but as independent digital entities managing resources and services at machine speed. Mike Hicks, Chief Social Navigation Analyst at Cisco ThousandEyes. Agent AI is changing the fundamental requirements of the network ecosystem. Traditional network testing that focuses on throughput and latency is becoming inadequate. You might like it. Instead, networks must provide real-time agility, contextual intelligence, and advanced resilience to support unexpected computer-controlled activity patterns. The primary web user is no longer the person opening a browser tab; Today’s infrastructure must manage autonomous agents that execute hundreds of API calls and complex, interdependent workflows in seconds. For providers, this marks a significant shift in performance expectations, requiring new approaches to visibility, orchestration, and service delivery optimization. Redefining network performance. Human-driven internet usage follows predictable patterns: Users manually navigate websites, open applications, and make relatively rare, sequential requests. Request Sign up for the TechRadar Pro newsletter to get all the top news, opinions, features and advice you need to make your business successful! Agent AI works differently by issuing rapid batches of API calls, aggregating data from multiple sources simultaneously, and performing multi-step processes with minimal supervision. For example, a single AI agent can query flight APIs, hotel databases, weather services, and payment systems – all at the same time. At the same time and within milliseconds. This shift allows networks to optimize for dozens of user connections supporting automated networks of orchestrated requests spanning multiple services and data centers. You might like it because the agent starts additional automated processes that can scale the volume and complexity of these interactions unexpectedly. This new reality requires adaptive monitoring capabilities that go beyond traditional performance metrics to include security audits and data integrity checks. As agents make autonomous decisions and transactions in third-party systems, networks must enable reliable handovers across domains and providers, maintaining accuracy and security regardless of the complexity or scale of automated processes. The workflow. The change in infrastructure occurs in two ways. These growing requirements are driving infrastructure changes in two directions simultaneously. Hyperscale cloud providers are extending their reach to provide backbone services for AI-driven applications, while at the same time the demands of agent-based AI for specialized computing resources, regulatory compliance, and significant power consumption are accelerating the growth of dedicated AI data centers and specialized infrastructure. Infrastructure evolution is leading to a mesh, distributed architecture in which data and computing flows across public clouds, private facilities, edge nodes, and IoT endpoints. An example of this trend is the growing number of specialized cloud GPU providers (sometimes called neo-clouds) that offer bare metal GPU-as-a-service optimized for AI workloads. These providers often target specific requirements that traditional cloud services cannot effectively address, such as specific hardware configurations, pricing models, or regulatory compliance requirements. Organizations can no longer rely on observing a few key dependencies; They need comprehensive visibility into constantly changing landscapes of service relationships and data flows. From passive transport to context-aware orchestration. This infrastructure transformation requires networks to evolve beyond simply moving data quickly. Networks must be active participants in service delivery, implementing policies that understand application context and business needs, rather than acting as passive transport layers. In an agent-based system, a single missed packet or degraded connection can cause cascading failures in automated workflows, disrupting business outcomes in ways that may not be immediately apparent to human operators. For example, if an AI agent managing supply chain logistics loses contact with a critical pricing API, it can lead to suboptimal purchasing decisions that worsen over time. Networks supporting agent-based AI must implement context-aware communication: ensuring quality of service based on application criticality, securing data flows across domain boundaries, and providing real-time visibility into agent workflows and service interactions. The measure of a network’s success can shift from whether data moves quickly to whether agents perform their tasks safely and efficiently according to business logic. Service chains must be visible and manageable, even if the structure of these chains is constantly changing depending on the decisions of agents and external conditions. Rebuilding networks for trust and intelligence Agent-based AI is fundamentally changing the demands of the internet infrastructure. The next generation of digital networks will be distinguished not only by their capacity or speed, but also by their ability to provide adaptive intelligence, as well as trust and transparency in a distributed, evolving service environment. As networks move from passive infrastructure to active orchestrators of digital value, providers that can deliver dynamic, resilient, and intelligent services will be positioned to support the emerging agent economy. Success will depend on building systems that can understand, adapt to, and manage the complex, automated interactions that will increasingly define how value moves across the internet. The transformation is already underway. The question now is what network design and service delivery methods will be most effective in supporting this new paradigm. We have launched the best business intelligence platform. This article was created as part of TechRadarPro’s expert insights channel, where we profile the best and brightest minds in today’s tech industry. The opinions expressed here are those of the author and do not necessarily reflect the views of TechRadarPro or Future plc. If you are interested in participating, learn more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
প্রকাশিত: 2025-10-21 20:23:00
উৎস: www.techradar.com
