Agents

The increasing adoption of large language models (LLMs) in enterprises has led to a rise in prompt injection attacks, which exploit the disconnect between assumptions about LLMs and their actual characteristics. According to the OWASP LLM Top 10 (2025), prompt injection is the most critical category of LLM-specific vulnerabilities, and CrowdStrike's 2026 Global Threat Report documented over 90 organizations affected by prompt injection attacks in 2025. These attacks have evolved to target multi-agent architecture, retrieval-augmented generation (RAG) pipelines, model routers, and long-term memory capabilities, making it essential for engineers to address this threat when deploying AI systems at scale. The practical implication for engineers is to develop strategies to mitigate prompt injection attacks and ensure the secure deployment of LLMs.

The engineering of reliable agentic workflows is a problem about variance, not speed, and requires counterintuitive fixes to deliver high-quality answers consistently and on time. Not mentioned are specific numbers, model names, or benchmark results. The practical implication for engineers building AI systems is that they need to focus on reducing variance to ensure reliable and timely delivery of answers.

This article provides a tutorial on setting up a production-ready local coding agent using open-source tools and open-weight large language models (LLMs). The local stack consists of a coding agent harness that uses a local model hosted through an inference engine/runtime server, allowing for transparent, inspectable, and cost-effective coding workflows. The author highlights the benefits of local solutions, including predictable costs, reproducibility, and offline use. The practical implication for engineers building AI systems is the ability to create custom, flexible, and cost-effective coding agents that can be tailored to specific needs.

The distinction between agentic workflows and autonomous agents lies in control flow ownership, with agentic workflows being human-driven and autonomous agents possessing self-directed control. While agentic workflows can leverage AI components, they do not independently execute tasks, whereas autonomous agents do. This dichotomy affects the level of human oversight required, with agentic workflows necessitating human intervention and autonomous agents operating with minimal human input. The choice between these approaches depends on the desired degree of autonomy and the complexity of the tasks being executed.

Anthropic's Claude Code has increased engineering productivity by roughly three times, shifting the bottleneck from coding to decision-making on what to build. This has led to a need for more product managers to define the product roadmap and prioritize features. The industry is undergoing a structural shift, where the engineer's role is evolving from solely writing code to also deciding what to build. The practical implication for engineers building AI systems is that they need to develop product thinking skills to remain relevant.

Researchers from MIT and Microsoft have developed an intelligent system that streamlines the process of designing agentic workflows, automatically optimizing the implementation and reducing computational units, energy requirements, and costs. The system allows developers to describe the desired workflow in plain language, without needing to specify all details in advance, and adjusts configurations on the fly based on user priorities. This approach has been shown to significantly cut energy requirements and costs compared to traditional approaches without hampering performance. The practical implication for engineers building AI systems is that they can now design and deploy more efficient agentic workflows, reducing waste and improving overall system performance.

The NVIDIA Agent Toolkit provides a foundation for building specialized AI agents that can be customized, controlled, and trusted by enterprises and developers. This toolkit includes models, tools, skills, and a secure runtime, enabling the creation of digital AI coworkers that can reason, use tools, and take action. With the NVIDIA Agent Toolkit, businesses can build specialized AI agents that fit their specific workflows, leading to increased efficiency and productivity. The practical implication for engineers building AI systems is that they can now create customized AI agents that can be integrated into existing systems and workflows.

Researchers at the National University of Singapore have developed MRAgent, a framework that enables AI agents to dynamically develop their memory based on accumulating evidence, reducing token consumption and runtime costs. MRAgent uses a "Cue-Tag-Content" mechanism to organize its database, allowing for efficient and scalable active exploration of memory. This approach overcomes the limitations of passive retrieval pipelines, which can fill the LLM's context window with noise and degrade reasoning. The framework uses 118K tokens per query, significantly less than other agentic memory management approaches like LangMem, which burns through 3.26M tokens. This reduction in token consumption has significant practical implications for engineers building AI systems, as it can lead to cost savings and improved performance.

Stripe built a production-grade AI agent system on AWS using Amazon Bedrock, reducing review handling time by 26 percent while maintaining human oversight and achieving over 96 percent helpfulness ratings. The system, based on Stripe's ReAct agent framework, utilizes task decomposition, orchestration patterns, and cost optimization through prompt caching to scale compliance operations. This approach addresses the $206 billion global compliance burden by identifying 95% of card-testing attacks in real time and reducing unnecessary customer friction by 20%. The practical implication for engineers building AI systems is the importance of designing agentic systems that balance automation with human oversight and accountability.

The AI landscape has shifted from models that simply know to agents that do, with foundation models being used as cognitive engines for AI agents in the physical world. To be useful in high-stakes physical environments, agents need to be grounded in physical laws and operational constraints, overcoming the challenge of hallucination. Four approaches to grounding AI agents are proposed, including physics-guided deep learning, which integrates first-principle physical knowledge into the foundation model in pretraining. This ensures that predictions obey governing physical laws, making agents physically consistent and operationally reliable. The practical implication for engineers building AI systems is that they must consider the physical constraints of the environment in which their agents will operate.

The 2026 Axonius Actionability Report reveals that 12.7% of devices in a 298,000-device median inventory are missing their expected security agent, resulting in incomplete data for autonomous security agents. This gap is critical as SOC and XDR vendors push more autonomous investigation and remediation into production, relying on the same dashboards and coverage percentages that human analysts have learned to work around. The report highlights the need for complete data to ensure effective security, with 63% of respondents stating that the underlying data lacks important information. This has significant implications for engineers building AI systems, as autonomous agents will treat incomplete data as ground truth and act on it at machine speed.

The article discusses building a powerful Large Language Model (LLM) knowledge base, suggesting the use of coding agents to power it. Not mentioned are specific numbers, model names, benchmark results, or architectural details. The practical implication for engineers building AI systems is the potential to leverage coding agents for knowledge base construction.

The authors propose a solution to transform legacy enterprise services into agents capable of participating in Agent-to-Agent (A2A) interactions using agentic overlays, a thin wrapper layer that exposes REST APIs as tools compatible with the Model Context Protocol (MCP). This approach allows enterprises to add A2A capabilities to existing REST services without rewriting business logic, duplicating code, or running parallel infrastructures. The agentic overlays enable autonomous agents to collaborate, reason, and coordinate through structured messaging, reducing agent sprawl in the infrastructure. The practical implication for engineers building AI systems is that they can leverage agentic overlays to integrate legacy services with A2A protocols, facilitating the adoption of AI in enterprise environments.

Seltz, an agentic infrastructure startup, has raised $12.5 million in seed funding to develop a more efficient infrastructure for AI agents to search the web. The funding round was led by Speedinvest and B Capital, with participation from several other investors. This investment aims to enhance the ability of AI agents to navigate the web and find relevant information. The practical implication for engineers building AI systems is the potential to leverage Seltz's infrastructure to improve the search capabilities of their AI agents.

The performance of AI agents is hindered by the intent-execution gap, which is the mismatch between what the model intends and what the harness executes. Minimizing this gap is sufficient to achieve state-of-the-art performance across diverse agentic benchmarks. The Simple Strands Agent (SSA) is introduced as a lightweight and customizable single-agent harness designed to close the gap between reported and actual performance. Effective agent design is not entirely model agnostic, and model-harness codesign is critical in achieving optimal performance. This has significant implications for engineers building AI systems, as it highlights the importance of considering the model-harness interface and identifying invariant components that remain effective across model upgrades and environments.

NVIDIA is bringing trusted, 24/7 AI agents to telecom operations, enabling autonomous networks and operations where AI agents proactively watch for problems and coordinate changes across network, IT, and business systems. The company is demonstrating the building blocks of a secure, telecom autonomy platform, including synthetic data, telecom-domain models, secure agent runtimes, and simulations. This platform allows agents to understand operator intent, act safely across business and network domains, and keep humans in control of policy. The practical implication for engineers building AI systems is the ability to create more autonomous, resilient networks and power richer AI-driven services for consumers and businesses.

OpenAI has announced a limited preview of its GPT-5.6 model series, consisting of three models: Sol, Terra, and Luna, with the flagship Sol model delivering a major performance gain for long-running coding, cybersecurity, and agentic tasks. The GPT-5.6 series introduces a new max reasoning effort mode and an ultra mode, which expands past the structural boundaries of a single standalone model, deploying specialized "subagents" to divide, conquer, and accelerate multi-step, long-horizon projects. The models have achieved state-of-the-art scores on various benchmarks, including Terminal-Bench 2.1 and Agent's Last Exam. The limited preview is available to a narrow set of trusted partners and organizations, with a broader public launch pending completion of a 30-day review process by the U.S. government. The practical implication for engineers building AI systems is that they will need to na

The article discusses building a lightweight research agent using various tools such as Gemma 4, Ollama, OpenAI Agents SDK, and Tavily MCP, enabling the transition from a local Large Language Model (LLM) to a tool-using agent. This integration allows for more complex tasks and improved performance. The practical implication for engineers building AI systems is the ability to leverage these tools to create more advanced and capable agents. The use of these specific tools and frameworks can streamline the development process and enhance the functionality of AI agents.

Researchers from MIT and other institutions have made a significant finding in the field of imperfect-information games, where two contestants compete in a zero-sum game. Their study shows that policy gradient methods, a general-purpose algorithm, can outperform specialized game-theoretic algorithms in certain situations. This challenges the long-held assumption that game-theoretic algorithms are superior in this setting. The researchers used neural networks to participate in imperfect-information games and found that policy gradient methods can work better than specialized algorithms. This has practical implications for engineers building AI systems that need to make decisions in complex, dynamic environments.

The Chaplin solution utilizes AI agents powered by Amazon Bedrock and exposed through the Model Context Protocol (MCP) to provide self-service health event analytics for AWS Health notifications. This approach enables teams to ask questions in natural language and receive precise, contextualized answers without relying on AWS Support. With Chaplin, teams can identify actionable health insights, prioritize events, and make informed decisions. The practical implication for engineers building AI systems is that they can leverage Chaplin to streamline health event management and focus on innovation rather than reactive firefighting.

The author benchmarked three approaches to multi-agent conversations: raw chat history, vector-only Retrieval-Augmented Generation (RAG), and a context graph layer. The results showed a weakness in relational retrieval, highlighting the need for a more comprehensive approach. The context graph layer was built to address this weakness, providing a more robust solution for multi-agent memory. This has significant implications for engineers building AI systems that require complex conversation management.

Building agentic AI applications on a modern data mesh strategy on AWS requires fine-grained access control enforced at every layer of the data interaction chain. The proposed architecture extends the original with three key changes: replacing Amazon OpenSearch Serverless with Amazon S3 Vectors, replacing general-purpose Amazon S3 with Amazon S3 Tables governed by AWS Lake Formation, and exposing the data mesh as Model Context Protocol (MCP) tools through AgentCore Gateway with AWS Lambda-backed interceptors. This approach provides a secure, scalable data foundation for production agentic AI, reducing vector storage and query costs by up to 90% and increasing transactions per second by up to 10 times. The practical implication for engineers building AI systems is the ability to enforce fine-grained access control and provide a governed data mesh for agentic AI applications.

A recent benchmark highlights the performance of GBDTs and agents in a payment-fraud detection scenario, focusing on latency, cost, and reproducibility. The results show that GBDTs excel in the hot path, while agents dominate the cold path. This distinction has significant implications for engineers designing AI systems for payment-fraud detection. The benchmark provides a reproducible framework for evaluating the effectiveness of different approaches. For engineers building AI systems, this means considering the strengths of both GBDTs and agents when designing payment-fraud detection pipelines.

By leveraging a multi-agent pipeline, the author achieved a 30% improvement in text-to-SQL query accuracy and a 25% reduction in latency compared to a single-agent approach. The pipeline consists of a language model, a SQL parser, and a query optimizer, which are integrated using a custom orchestration framework. This setup allows for more efficient handling of complex queries and better scalability. However, it also introduces additional complexity and requires careful tuning of each component.

Anthropic has introduced Claude Tag, a new version of its chatbot Claude, designed to operate within Slack as a virtual employee, assisting multiple employees with tasks for related projects. This build upon existing agentic AI tools, including Claude Code. The integration of Claude Tag into Slack enables it to work across entire organizations, enhancing collaboration and productivity. This development has practical implications for engineers building AI systems, particularly those focused on integrating AI tools into existing workflows and collaboration platforms.

Nvidia is betting on agentic AI to accelerate biotech discovery, as announced at the Bio International Convention in San Diego. The company's vice president and general manager of healthcare and life sciences, Kimberly Powell, made the case for agentic AI in a special address. Not mentioned are specific numbers, model names, or benchmark results. The practical implication for engineers building AI systems is the potential application of agentic AI in biotech discovery. Agentic AI may enable more efficient and effective discovery processes.

This article presents a technical guide on building a protein research copilot using Amazon Bedrock AgentCore, which enables researchers to search for structurally similar peptides across large datasets using natural language queries. The system combines natural language query parsing, vector similarity search over protein embeddings, and AI-generated scientific summaries of search results. The copilot is built using the Strands Agents SDK and deployed to Amazon Bedrock AgentCore for production serving. The practical implication for engineers building AI systems is the ability to create conversational interfaces that can handle complex research workflows and provide accurate results.

ClickHouse, a column-store database management system, has been integrated with agentic AI to provide real-time analytics, enabling millisecond responses for AI agents to make decisions and access data rapidly. This integration has improved the performance of AI agents by 300% compared to traditional batch-oriented systems. Real-time analytics have become essential for AI agents to function effectively, and ClickHouse has filled this need by providing a scalable and high-performance data layer. The success of this integration has paved the way for the widespread adoption of AI agents in various enterprise applications.

Ampersend has built a pay-per-intelligence routing layer on top of Amazon Bedrock AgentCore Payments, enabling AI agents to autonomously route tasks to the most effective model and pay per request within governed limits. The two-hop payment pattern allows agents to pay for intelligence services across multiple model providers through a single integration point, powered by the x402 open protocol. This solution addresses the infrastructure gap in payment infrastructure for autonomous agents, providing a managed payment infrastructure that is secure, auditable, and governed. The practical implication for engineers building AI systems is that they can now focus on agent logic without having to build bespoke billing integrations, credential management, and payment orchestration from scratch.

Amazon Bedrock AgentCore now offers a fully managed web search capability, allowing AI agents to access up-to-date information from the web without infrastructure overhead. This feature, compatible with the Model Context Protocol (MCP), provides a purpose-built web index spanning tens of billions of documents, updated continually to reflect new content within minutes. The privacy model ensures that queries stay within AWS, and retrieval can combine a knowledge graph with semantic snippet extraction. This development has significant implications for engineers building AI systems, as it addresses the limitation of frozen knowledge at training time and enables agents to respond to real-time queries.

NVIDIA XR AI is now available in public beta, providing a framework for developers to build multimodal AI agents for AR glasses and XR devices. This framework enables the creation of agents that can interact with users in a hands-free manner. Not mentioned are specific numbers, model names, or benchmark results. The practical implication for engineers building AI systems is the ability to create more interactive and immersive experiences for users. The public beta release of NVIDIA XR AI allows developers to start building and testing their own multimodal AI agents.

The HPE AI Factory with NVIDIA is expanding to support the increasing adoption of agentic AI, integrating NVIDIA Vera CPU and NV Switch for accelerated model inference and training, aiming to reduce latency and improve scalability for enterprise AI workloads. This expansion enables enterprises to move agentic AI from proof of concept to production, with a focus on multi-step AI agent pipelines. The updated HPE AI Factory is designed to handle the complex computations required for agent-based AI, with a scalable and flexible architecture that can support a wide range of AI workloads. This expansion is a significant step towards making agentic AI more accessible and practical for enterprises.