A significant shift is underway in the AI landscape, with major tech companies pivoting toward agent-based systems capable of performing complex tasks with minimal human input. This week, both Amazon and Salesforce made major moves into agentic AI, while Gartner identified AI agents as the top technology trend for 2025.

Agents become the new AI frontier

Amazon has established a dedicated agentic AI group under AWS executive Swami Sivasubramanian, positioning autonomous AI systems as their next multi-billion-dollar business opportunity. This strategic move aims to develop AI that can operate independently without constant user input, aligning with upcoming enhancements to Alexa. AWS is undergoing broader restructuring to support this new focus.

Meanwhile, Salesforce launched AgentExchange, a marketplace for AI agents that includes pre-built agents, templates, and actions across various industries. The platform features assets from 200 initial partners including Box, DocuSign, and Workday, with full functionality expected by spring 2025.

Tools emerging to support agent development

As agent-based AI gains momentum, new tools are emerging to support developers building these systems. Arch Gateway, created by Envoy Proxy contributors, provides a dedicated middleware layer that centralizes routing, security, and observability functions for AI applications. This proxy server addresses complex infrastructure challenges, allowing developers to focus on building intelligent agents rather than managing infrastructure.

In the academic sphere, Scholium has launched an AI research assistant that demonstrates agents' practical applications. The tool streamlines academic research by finding relevant papers, generating summaries, and creating citations from the arXiv database, with plans to expand to PubMed and major academic journals.

Data-centric approach drives business transformation

Gartner's 2025 data trends highlight this shift toward AI-centric technologies and ubiquitous data analytics. The report places AI agents at the top of recommended technologies, emphasizing composite AI approaches that combine multiple techniques, small language models for specific applications, and synthetic data for enhancing information protection.

Companies must consider:

We publish daily research, playbooks, and deep industry data breakdowns. Learn More Here