Automation
AWS Marketplace Deploys AI Agents for Autonomous Software Buying
Amazon's new procurement tools let companies authorize agents to purchase and deploy software independently, with configurable approval thresholds and integrated security reviews.
Key takeaways
- AWS Marketplace now allows companies to authorize AI agents to purchase software autonomously within defined budgets and approval rules.
- Integrated security review tools from Drata and Vanta can reduce vendor compliance checks from weeks to minutes by automating documentation verification.
- Buyer search behavior has shifted from keyword queries to natural language descriptions of project requirements and functional needs.
- Organizations can configure different authorization levels, from requiring human approval on all purchases to granting full autonomy for specific product categories.
- The system addresses procurement delays that have become critical bottlenecks as AI tools evolve rapidly and companies deploy more operational agents.
AI agents get procurement authority on AWS Marketplace
Amazon Web Services has launched a suite of AI agents designed to handle software procurement end-to-end on its AWS Marketplace platform. The system allows companies to pre-authorize agents to make purchases within defined parameters, eliminating human approval bottlenecks for routine software acquisitions.
The new capabilities span the entire procurement lifecycle. Agents can search for products, evaluate contract terms, verify vendor security credentials, complete purchases, and configure software deployments. Organizations set spending limits and establish rules determining which transactions require human review.
Matt Yanchyshyn, vice president of AWS Marketplace and partner services, described a scenario where an agent building a company website needs to integrate third-party authentication. Previously, the agent would pause for approval before purchasing a service like Okta. Under the new system, the agent can initiate a free trial or pay-as-you-go subscription autonomously, then build and test a working prototype without delay.
How buyer behavior is shifting
AWS has observed fundamental changes in how customers search for software. Rather than keyword searches like "database," buyers now describe their project goals and functional requirements in natural language. AI agents translate these descriptions into detailed specification lists, which Marketplace uses to recommend appropriate software and service combinations.
The platform has added a checkout assistant that answers questions about pricing structures, licensing terms, and purchase order requirements. AWS is also introducing a data category on Marketplace, recognizing that agents increasingly need access to datasets—for instance, an agent evaluating startup investments could pull funding information from Crunchbase.
Security reviews accelerated through automation
While agents can execute purchases independently, security validation remains essential. AWS has integrated compliance automation providers Drata and Vanta directly into Marketplace to streamline vendor security assessments. According to a June 2026 G2 survey, 39% of B2B software buyers identified IT security reviews as the primary delay after vendor selection.
Adam Markowitz, CEO of Drata, explained that a typical security review involves checking 50 to 60 requirements derived from extensive questionnaires. AI vendors introduce additional criteria around data handling and model governance. Agents can now retrieve vendor documentation such as SOC 2 reports, verify compliance requirement by requirement, and cite specific passages supporting their conclusions—reducing review cycles from weeks to minutes.
Configurable autonomy for different deal sizes
Yanchyshyn emphasized that companies retain full control over agent authorization levels. Organizations can require human approval for every transaction or grant agents complete independence for specific product categories. The imperative, he said, is speed: AI tools evolve rapidly, and companies cannot afford three-month procurement cycles when capabilities change week to week.
Bill Fitzgerald, vice president at Vanta, expects agents to handle high-volume, low-value purchases fully autonomously while major contracts—seven, eight, and nine-figure deals—will continue requiring human oversight.
Why it matters
This shift addresses a critical friction point in enterprise AI adoption. As organizations deploy more AI agents for operational tasks, procurement delays have become a significant constraint on agent effectiveness. By allowing agents to acquire necessary tools and services within governed parameters, companies can maintain deployment velocity without sacrificing financial controls or security standards. The integration of automated security reviews is particularly significant, as compliance verification has historically created the longest delays in software procurement cycles.
These details were first reported by Fortune.