Ddos 
As AWS re:Invent 2025 unfolds, the company has unveiled a wide array of new technological collaborations, spanning AI deployments across video understanding, financial payments, transportation, smart buildings, and other vertical domains.
Among the highlights, TwelveLabs introduced what it calls the “world’s most powerful” video-understanding model, Marengo 3.0, on the Amazon Bedrock platform. Meanwhile, Visa announced a partnership with AWS to advance Agentic Payments, enabling AI services to securely conduct transactions on behalf of users.
TwelveLabs’ Marengo 3.0 addresses a long-standing enterprise challenge: the inability to effectively utilize unstructured video data.
Unlike conventional models that analyze footage frame by frame, Marengo 3.0 treats video as an integrated dynamic system, capable of directly linking dialogue, gestures, actions, and emotional cues. The company claims the model can reduce storage costs by 50%, double indexing performance, support videos up to four hours in length, and recognize 36 languages—allowing businesses to search and understand video content as easily as searching text.
In the financial sector, Visa has announced a collaboration with AWS that merges its global payments infrastructure with AWS’s AI technologies to realize secure Agentic Payments.
The two companies will offer open blueprints on Amazon Bedrock AgentCore to help developers build intelligent workflows. Rubail Birwadker, Visa’s Senior Vice President of Global Growth, described the initiative as the emerging “trust layer” of the agent economy.
For example, a user might instruct an AI agent: “If basketball tickets drop below $150, buy them for me.” The agent would then autonomously compare prices, place the order, and complete the checkout process. Early partners include services such as Expedia and Intuit.
On the software and data front, AWS announced a partnership with S&P Global to integrate its trusted market, financial, and energy intelligence directly into customers’ AI workflows.
The key enabler is the Model Context Protocol (MCP). Through two newly integrated MCP servers, S&P Global’s data can be seamlessly accessed from Amazon Quick Suite. This means that AI agents built by enterprise users will not only converse but also handle sophisticated financial analysis.
For instance, using the S&P Global MCP for Kensho—an LLM-enabled API—AI agents can retrieve Capital IQ financials or earnings-call transcripts, or access real-time energy market insights through the S&P Global AI Ready Data MCP Server.
Bhavesh Dayalji, S&P Global’s Chief AI Officer, emphasized that this ensures customers can always access trusted data—whether in cloud platforms, LLMs, or agentic systems—thus optimizing decision-making across the board.
Additional collaboration examples include:
- Lyft: Using Anthropic’s Claude model and Amazon Bedrock to build an Intent Agent. When a driver asks, “Why isn’t my earnings total showing?”, the AI can directly query backend data—discovering, for example, that the driver has just completed three rides—and quickly resolve the issue. This has reduced Lyft’s average customer-service resolution time by 87%.
- Nissan: Launching the AWS-based Nissan Scalable Open Software Platform, accelerating software-defined vehicle (SDV) testing by 75% and enabling collaboration among 5,000 developers worldwide. Nissan also revealed plans to integrate a more advanced ProPILOT AI system by 2027.
- Trane Technologies: Through its BrainBox AI system, the company optimized HVAC performance at three Amazon Grocery pilot sites in North America, cutting energy usage by nearly 15%. The program is expected to expand to additional Amazon Fresh locations in 2026.
- BlackRock: Announcing that its investment management platform, Aladdin, will migrate to AWS, providing clients with enhanced flexibility in portfolio decision-making. U.S. availability is expected in the second half of 2026.
Donate