OpenAI Adds Fuel to the AWS Engine

Research evidence excerpt

OpenAI Adds Fuel to the AWS Engine

ing everywhere, and security is a

key reason why. A senior solutions architect at a cloud data platform vendor, who runs 20-30

agents himself with an orchestrator that routes tasks to the most suitable model, emphasized

that while agents can run continuously, proper guardrails must be enforced upfront or they risk

producing false negative or false positive outputs. Compliance rules still need to be applied, and

legacy environments add further complexity. The takeaway is that lack of security and governance

could become a meaningful headwind to agent proliferation and, more broadly, a risk to enterprise

AI adoption.

Kiro traction & AgentCore maturity. Kiro (AWS's agentic IDE) is gaining real traction among

practitioners. A solution architect at one of AWS's Premier-tier services partners described Kiro

as more thorough than alternatives. Kiro's spec-based programming approach produces three

structured docs: a requirement doc (user stories and acceptance criteria), a design doc (technical

architecture, data flow), and a tasks doc (discrete, trackable implementation steps), whereas Claude

and others typically run a single spec and execute immediately. This aligns with AWS's push to build

production-grade enterprise apps from the start. On AgentCore (AgentCore agent tasks have grown

15x in 6 mos), he noted the new harness lets developers orchestrate their own logic, agents hosted

on AgentCore are more cost-efficient than those hosted on Bedrock directly, & built-in guardrails

are a real advantage. He also viewed AgentCore as more mature than GOOGL's Vertex AI.

Brent Thill * | Equity Analyst

AWS Marketplace continues to be a strategic retention tool. The common thread across our +1 (415) 229-1559 | bthill@jefferies.com