AWS re:Invent 2025 - Agentic AI: The Next Frontier of Cloud Intelligence (AIM342)

Summary of AWS re:Invent 2025 - Agentic AI: The Next Frontier of Cloud Intelligence (AIM342)

Overview of ATOSS and the Agentic AI Landscape

• ATOSS is a large global consulting firm with a growing data and AI business line, aiming to reach 10,000 employees by 2028.
• The presentation aims to "demystify" agentic AI and provide an honest assessment of the current state of AI productization.
• The Gartner Hype Cycle is used to illustrate the current state of AI, with agentic AI at the "peak of inflated expectations" and model ops in the "trough of disillusionment".
• The key challenges highlighted include data availability, governance, and the ability to prioritize and productize AI use cases - with only 5% of planned use cases actually making it to production.
• The presenters categorize the AI landscape into 5 segments, from consumer-focused "easy entry" AI to the "market makers" doing large-scale investments in advanced AI.

ATOSS' Approach to Agentic AI Adoption

• ATOSS focuses on category 3 and 4 AI use cases, which require significant partnerships and technology to implement but offer high business value.
• A key challenge is user adoption, with some users "shadow AI" or abandoning AI tools entirely due to poor performance or fit.
• ATOSS aims to address this through a combination of consulting, technology, and building a community around agentic AI.

ATOSS Digital Advisor: An Agentic AI-Powered Cloud Advisory Service

• ATOSS has developed a "Digital Advisor" platform that leverages agentic AI to modernize their cloud advisory services.
• The platform uses a combination of AWS services, including API Gateway, Lambda, Step Functions, and the Amazon Bedrock agentic AI platform.
• It automates the initial information gathering and assessment process, allowing advisors to have more meaningful discussions with customers about their goals and priorities.
• The agentic AI agents handle tasks like web research, querying the customer's AWS environment, and generating personalized recommendations and maturity assessments.
• This approach aims to improve the efficiency and value of the cloud advisory process for ATOSS customers.

Key Takeaways and Opportunities

• Agentic AI is rapidly evolving, with new capabilities emerging around web browsing, multi-step task execution, and strong governance.
• ATOSS is focused on building frameworks, deployment platforms, and orchestration tools to help clients effectively leverage agentic AI.
• The goal is to enable more personalized, self-adaptive AI agents that can leverage real-time enterprise data to drive better business outcomes with minimal human intervention.
• ATOSS sees opportunities in industry-specific "super agents", autonomous and multimodal AI capabilities, and ensuring robust safety, alignment, and oversight.
• Successful adoption of agentic AI will require an "AI-first" strategy, vertical industry solutions, and a focus on data-first architecture and AI agent orchestration.