Amazon Web Services (AWS)

AWS CEO Matt Garman made agentic AI AWS's next multi-billion-dollar bet when most of the market was still getting familiar with the concept. That early conviction is paying off. In Boston Consulting Group (BCG)'s CEO Moments of Truth, Matt tells BCG CEO, Christoph Schweizer how AWS reorganized around agents, putting top leaders single-threaded on the opportunity and rearchitecting how teams build from the inside out. What used to take hundreds of people to build now takes a handful. Teams across AWS are seeing 5-10x output gains, and customers across industries from banking to drug discovery are leaning in to transform their businesses. Watch the full conversation to hear Matt's advice for CEOs on their own AI transformation journeys. https://go.aws/3Uv1nvU.

AI agents can move money, change infrastructure, and ship code. Giving them more autonomy requires confidence in the boundaries around that work. An agent may take different paths through the same task. Testing remains essential, but passing tests alone doesn’t establish that every future action will stay within policy. Reviewing every step by hand doesn’t scale. Teams need controls that check actions against explicit rules and bring people in for decisions that need judgment. Trust in Depth from AWS organizes that work around four questions: • Scope: What is the agent for, what capabilities does it need, and what must it never do? • Know: Who authorized it, and what identity does it carry? • Bound: Do its actions stay within the limits you set, and can it still do the job? • Observe: Can you reconstruct its actions and learn from the outcome? Automated reasoning strengthens these layers with mathematical proof for specific rules under defined assumptions. Decisions that depend on context and judgment still need evaluation and human oversight. Together, these layers help organizations expand agent autonomy as evidence builds. Read the full post from Shawn Bice, VP, Automated Reasoning Group, AWS. https://go.aws/3UsRdvK

Au moins 30 % des projets d’IA sont abandonnés après la preuve de concept. AWS peut faire en sorte que cela se passe différemment pour vous.

En seulement six étapes, vous pouvez créer et mettre à l’échelle en toute confiance des applications et des agents d’IA, du prototype à la production.

How do you search 65 petabytes of video data? March Networks solved the enterprise video storage challenge by combining AWS cloud solutions with generative AI. Using Amazon Bedrock & Amazon S3, they're helping Fortune 500 companies turn massive video archives into searchable intelligence. Learn more. https://go.aws/4ellnI4

Beaucoup d’entreprises ont du mal à faire entrer en production leurs prototypes d’IA. Ce guide peut facilement vous aider à créer et mettre à l’échelle.