AWS re:Invent 2025: Live Updates on New AI Innovations and More

AWS re:Invent 2025 delivered game-changing announcements, AI agents that solve customer issues in minutes, multicloud simplicity with Google Cloud, SDV acceleration with Nissan, and powerful new models like Marengo 3.0. These innovations aren’t upgrades; they’re a blueprint for how enterprises will operate, innovate, and compete in the AI-accelerated decade ahead.

The AWS re:Invent 2025 conference has once again become the epicentre for cloud computing, AI innovations, and cutting-edge technology developments. 

As the world's leading provider of Cloud Computing Infrastructure, Amazon Web Services (AWS), used this event to announce the launch of many new technology products, partnerships, and advancements that will transform the future of enterprise computing,AI, and Customer Experience. 

The announcements released at re:Invent 2025 clearly illustrate AWS's steadfast dedication to exploring new limits through agentic AI, multi-cloud networking, AI-Powered Sustainability, etc.

AWS Transform: Revolutionising Legacy Code Modernisation with Agentic AI

Many companies today struggle with out-of-date technology and the burden of maintaining "tech debt", a slow, complicated, manual process that consumes precious time and resources while stifling their ability to innovate. To help businesses alleviate these challenges, AWS has developed AWS Transform, an advanced new service that enables companies to update their legacy software and applications as much as five times faster than through traditional methods.

With the introduction of agentic artificial intelligence (AI) capabilities integrated into AWS Transform, organisations can efficiently and rapidly upgrade any type of software system, regardless of the programming language or complexity of the system.

Prior to the introduction of AWS Transform, updating legacy systems typically required extensive manual labour and took months or even years to complete; in contrast, AWS Transform greatly eliminates the amount of manual labour, and many organisations experience an average decrease in the time and resources spent on maintaining these systems by approximately 70%.

Currently, this service supports the entire Windows stack, including modern .NET applications, Microsoft SQL Server, and user interface frameworks, allowing companies to not only save significant time during their migration but to also significantly reduce their associated operational costs.

For example, Air Canada was able to use AWS Transform to migrate thousands of Lambda functions in only days, thereby achieving an 80% reduction in the time and costs involved in migrating to the cloud using traditional manual methods. The improvements made to AWS Transform are clear indications of AWS’s commitment to providing businesses with the tools they need to focus on innovation rather than continuing to struggle with outdated systems.

TwelveLabs and Marengo 3.0: breaking new ground in video understanding

TwelveLabs launched the Marengo 3.0 model, a revolutionary foundation model for video understanding, at AWS re:Invent 2025. Video has been the leading type of digital data, contributing 90% to digitised data. Up to this point, however, the tremendous complexity of video has made it impossible for businesses to extract any meaningful information from that data using existing AI models, as the models cannot interpret the shifting and intricate movement of video content.

Marengo 3.0 defeats this barrier by understanding video through the entire flow of time as a whole, pairing dialogue, gesture, motion, and emotion all together, and gives businesses the opportunity to explore and understand video content with a human level of understanding and comprehension on scale.

The model immediately cuts 50% of the cost for storage, doubles indexing, and is used to provide detailed video analysis for sports, both live and recorded, and offers businesses access to over 36 languages for video content up to four hours long. Jae Lee, CEO of TwelveLabs, emphasised, "Video represents 90% of digitised data, but that data has been largely unusable. Marengo 3.0 shatters the limits of what is possible."

This Amazon Bedrock offering is the first Video Model of its kind provided by any cloud provider, allowing companies to securely deploy the model to their own AWS Infrastructure.

Trane Technologies and AWS: Leading the charge in decarbonising Amazon’s grocery fulfilment operations

Trane Technologies and AWS have collaborated to improve energy efficiency while minimising carbon footprints through the use of BrainBox AI. By leveraging this innovative AI technology, Amazon has increased its energy savings in its grocery fulfilment centres by nearly 15% compared to previously established targets.

The partnership enables Amazon to further its efforts to achieve a net-zero carbon footprint by 2040 as detailed in The Climate Pledge. Using BrainBox AI, optimising controls for HVAC systems for Amazon's grocery fulfilment sites allows for self-regulated temperature control and energy optimisation to conserve resources, improving operational efficiencies.

Riaz Raihan, SVP and Chief Digital Officer at Trane Technologies, noted, “At Trane Technologies, sustainability is at the core of everything we do. This strategic collaboration demonstrates how sustainable solutions can drive strong returns while benefiting the planet.” 

Following the success of this initiative, further deployment is planned for Amazon’s other facilities, including distribution centres, with pilots set to begin in grocery stores by 2026.

Lyft and AWS: Reimagining Rideshare Customer Service with Agentic AI

One of the most anticipated developments at AWS re:Invent 2025 is a collaboration with Lyft, AWS, and Anthropic. This multisensory agentical AI will revolutionise the way customers interact with rideshare providers by creating an intent-based agent to help Lyft's 1 million drivers efficiently resolve driver-related problems as quickly as possible. Lyft developed an intent agent to provide timely solutions through a unique process that combines utilising context-based information with utilising uninterpreted data.

For example, if a customer poses the question, "Where is my money?" The AI agent will immediately retrieve all of that customer's most recent activity to provide an answer. This innovative solution offers a more than 87% decrease in the amount of time needed to resolve customer inquiries, and more than half of all customer inquiries can be resolved in less than three minutes.

Lyft's Ameena Gill, VP of Safety and Customer Care, stated, “The future of customer service has fundamentally shifted through AI.” This innovation represents a bold step toward more personalised, efficient, and proactive customer service experiences.

Nissan and AWS: Driving the future of software-defined vehicles

Automotive technology is evolving quickly, and Nissan is at the forefront of this evolution with the launch of its Scalable Open Software Platform (N-SOSP), which is built on AWS cloud services. N-SOSP allows Nissan to develop SDV technologies more quickly than ever before because of the ease of use it provides for developers working remotely. 

For example, through N-SOSP, Nissan can now perform testing on an average of 75 per cent quicker, while still enabling development in a shared environment for more than 5,000 global developers engaged in both software and hardware for SDVs.

As a result, Nissan can now deliver innovative vehicle features more quickly than ever before, setting the stage for future AI advancements like ProPILOT, an autonomous driving system. 

Kazuma Sugimoto, General Manager at Nissan, said, “Software development for SDVs is an extremely important strategy for Nissan.” With AI capabilities integrated into their software systems, Nissan is set to redefine the future of driving.

Visa and AWS: Enabling secure agentic payments

The Visa and Amazon Web Services (AWS) partnership has since borne Visa Intelligent Commerce, which allows Artificial Intelligence (AI) agents to carry out secure transactions independently. Through combining Visa's extensive payment system with AWS's AI capabilities, they have created a more user-friendly method of completing transactions, while providing an enhanced level of security and efficiency. 

In addition, Amazon Bedrock has shared the "open" blueprints of their AgentCore repository to support developers in creating intelligent AI workflows for various retail shopping, travel booking, and payment reconciliation tasks, facilitating the ability for AI agents to manage all aspects of today's complicated transaction processes. 

As a result, now AI agents can create product discovery, effect price comparisons, and execute secure checkouts with the ability to automate tasks.

Rubail Birwadker, SVP of Visa, said, "Visa Intelligent Commerce is designed to be the trust layer for the agent economy." This innovative system is reshaping how consumers and businesses interact with the digital commerce landscape.

BlackRock and AWS: Scaling aladdin for the future of investment management

The launch of Aladdin on AWS allows clients access to advanced risk modelling capabilities and robust enterprise-level analytical capabilities. Through Aladdin's expansion to AWS, clients are provided the means to make informed investment decisions more quickly, securely and efficiently.

Sudhir Nair, Senior Managing Director at BlackRock, stated, “By expanding Aladdin to AWS, we are giving clients more choice in where and how they deploy their technology ecosystem.” 

This collaboration allows BlackRock clients to tap into AWS’s capabilities, ensuring their investment decisions are based on the most up-to-date data and the most robust analytical tools available.

Amazon Connect: Enhancing customer service with AI-powered assistance

Amazon Connect sets new standards for customer service with its innovative capabilities for AI agents that enhance customer interactions. With the ability for Amazon Connect’s AI agents to interpret sentiment, context, and customer history, customers are able to benefit from natural and seamless conversations. 

Companies can now take advantage of Amazon Connect’s capabilities to automate both routine and complex tasks, enhancing the customer experience and reducing operational costs across voice & messaging channels.

Amazon Connect now also allows companies to benefit from Nova Sonic’s advanced speech models, which allows for human sounding conversations in numerous different languages and accents. 

In addition, through the use of AI-powered recommendations, businesses can provide personalised product recommendations during customer interactions. Consequently, businesses are transforming their method for engaging customers into a much easier and more personal experience.

AWS interconnect: Simplifying multicloud networking

AWS has partnered with Google Cloud to create a service called AWS Interconnect - Multicloud which simplifies the management of multiple cloud platforms: AWS and Google Cloud. 

With AWS Interconnect, Multicloud, businesses are offered fast and secure ways to connect private high-bandwidth links between the two providers. The combination of these services creates a seamless transition for organisations using both cloud platforms and removes the headaches associated with managing multiple clouds.

In addition to providing an effective method for establishing secure, private high-speed connections between AWS and Google Cloud, AWS Interconnect - Multicloud provides companies with an easy way to control, monitor, and manage their multicloud environments with confidence.

Deepgram and AWS: Empowering speech AI for real-time voice interactions

Deepgram has integrated its enterprise-level speech technology into Amazon SageMaker AI, Amazon Connect and Amazon Lex, providing streaming speech to text, text to speech and voice agent capabilities on AWS. Businesses can develop voice-powered applications that deliver real time, context based responses using the capabilities of the two integrated platforms; this will create improved experiences for customers.

Scott Stephenson, CEO of Deepgram, noted, “By bringing our streaming speech models directly into SageMaker, enterprises can deploy speech-to-text, text-to-speech, and voice agent capabilities with sub-second latency.”

This integration represents a significant leap forward in voice AI, offering businesses more control and flexibility in creating real-time, context-sensitive voice applications.