Foresight Ventures says AI agents are becoming part of online commerce systems. In its latest report, the firm explained how companies are building systems that allow AI agents to handle payments, transactions, and automated tasks. The next stage of AI growth may depend on execution systems instead of chatbot features alone.
AI Agents Are Taking on More Online Tasks
In a report released on May 8, Foresight Ventures reported that AI agents are moving beyond simple chat tools and starting to complete real-world tasks.
The report describes “agentic commerce” as a system where AI agents can search for products, compare prices, make payments, and complete transactions on behalf of users.
This shift is also changing how companies design software. In the past, software was mainly built for human users. Now, firms are increasingly creating systems that AI agents can directly read and interact with.
Foresight Ventures said AI agents depend more on APIs, permissions, and structured data than on visual interfaces or app layouts.
Major Companies Expand AI Commerce Infrastructure
In presenting use cases, the report highlighted several major companies developing systems for AI-powered commerce.
Among the list is OpenAI, which is developing tools that help businesses build and deploy AI agents. Google is also expanding AI tools for automated workflows and search.
In the payments sector, Stripe, Visa, and Mastercard are developing systems that could allow AI agents to make secure online payments.
The report also mentioned Coinbase and its x402 project, which aims to simplify online payments for AI agents and internet services. Foresight Ventures said these companies are now competing to become the infrastructure layer for AI-driven commerce.
The Report Focuses on “Agent Harness”
A major focus of the report is the concept of “Agent Harness,” which Foresight Ventures described as the framework that allows AI agents to safely operate in real-world environments.
The harness layer includes permissions, workflow tools, monitoring systems, risk controls, and feedback mechanisms.
According to Foresight Ventures, companies now care more about whether AI agents can complete tasks safely and reliably than whether they can simply answer questions.
The firm added that AI systems must also track actions, recover from errors, and improve from feedback over time.
Execution Data Could Become More Valuable
Notably, Foresight Ventures said that future AI competition may focus on execution data rather than model size alone.
Accordingly, the report explained that companies that collect data on AI actions, decisions, and results may gain stronger long-term advantages.
It also noted that industries such as finance, healthcare, cybersecurity, and legal services may adopt AI agents faster because their workflows are structured and results are easier to measure.
The report concluded that the next stage of AI development may depend on systems that help AI agents safely complete real-world work instead of simply producing better responses.
