Рубрика: AI

CognitiveLab has launched NetraEmbed, a new multimodal multilingual document retrieval model that supports 22 languages and the company claims delivers about 150% improvement over existing baselines.

The announcement was shared on December 8, along with the release of NayanaIR, an open source multilingual benchmark, and a preprint of the supporting research paper titled “M3DR: Towards Universal Multilingual Multimodal Document Retrieval”.

Adithya S Kolkavi, founder of CognitiveLab, said on X, “We are a Small Research Lab based out of India and we just dropped a one of a kind SoTA multimodal multilingual document retrieval model.”

The company in the blog post said NetraEmbed scores 0.716 on cross lingual retrieval tasks, up from the previous best of 0.284. It also records 0.738 on monolingual search. The model processes documents as images rather than relying on OCR, which helps it preserve charts, tables, diagrams, and layout.

It supports 22 languages covering English, Spanish, French, German, Italian, Hindi, Marathi, Sanskrit, Kannada, Telugu, Tamil, Malayalam, Chinese, Japanese, Korean, Arabic, Bengali, Gujarati, Odia, Punjabi, Russian, and Thai.

CognitiveLab said the model brings cross lingual document search from barely functional to production ready.

CognitiveLab also introduced ColNetraEmbed, a multi-vector variant that offers token level explanations. NetraEmbed uses compact embeddings at about 10 KB per document, compared to about 2.5 MB in traditional systems, enabling large scale indexing for enterprises.

The model offers flexible embedding sizes at 768, 1536, and 2560 dimensions without retraining.

The NayanaIR benchmark covers 23 datasets with nearly 28000 document images and more than 5400 queries and is designed for both monolingual and cross lingual evaluation.

The launch is part of CognitiveLab’s Nayana initiative focused on multilingual and multimodal document intelligence. Future models under the initiative will move beyond retrieval into deeper understanding and question answering across languages.

Both models are available on Hugging Face.

The post CognitiveLab Unveils NetraEmbed With 22 Languages & 150% Jump in Document Accuracy appeared first on Analytics India Magazine.

IBM is close to finalising a roughly $11 billion acquisition of data-infrastructure firm Confluent, according to a Wall Street Journal report, in what would be the tech giant’s biggest bet yet to strengthen its cloud software and AI ambitions.

The deal, which could be announced on December 8, the report noted, would give IBM control of Confluent, a widely adopted real-time data streaming platform that enables organisations to build, manage, and operate real-time data streaming applications at scale. Built on the open-source streaming platform Apache Kafka, companies across sectors use Confluent’s platform to process everything from bank transactions to website clicks.

The company’s latest offering, Confluent Intelligence, aims to help enterprises build and scale context-rich and real-time AI systems.

AIM reached out to Confluent but did not receive a response at the time of publishing.

The potential takeover comes at a pivotal moment for IBM, which is under investor pressure to reinvigorate growth in its cloud software division. Despite IBM posting a 9% year-on-year growth in its topline in its third quarter, its stock slipped 6% on extended trading following the earnings results as investors grew wary about slowing momentum in core cloud offerings, raising concerns over its long-term trajectory.

Confluent, valued at about $8.09 billion, has been exploring a sale and has hired an investment bank to manage the process after receiving interest from potential buyers, according to a Reuters report.

IBM, meanwhile, holds a market capitalisation of roughly $287.84 billion.

A successful acquisition would extend IBM’s M&A strategy under CEO Arvind Krishna, who has sharpened the company’s focus on cloud and software.

Last year, IBM acquired HashiCorp in a $6.4-billion deal aimed at expanding its cloud-native and automation capabilities amid rising enterprise spending on AI.

The interest in Confluent underlines the accelerating demand for data infrastructure platforms as companies race to build and deploy generative AI systems. In a similar move earlier this year, Salesforce agreed to buy Informatica for about $8 billion to enhance its AI-driven data stack.

The post IBM in Talks to Buy Confluent for $11 Bn in Cloud, AI Push: Report appeared first on Analytics India Magazine.

HCLTech has partnered with French semiconductor company Dolphin Semiconductor to co-develop energy-efficient chips targeting Internet of Things (IoT) and data centre applications, the company said in a statement.
Dolphin Semiconductor provides mixed-signal semiconductor intellectual property (IP) solutions for sectors including power management, audio, power metering, and design safety, serving industrial, high-performance computing, consumer electronics, IoT, and automotive markets.

Under the partnership, HCLTech will integrate Dolphin Semiconductor’s low-power semiconductor IPs into its system-on-chip (SoC) design and development workflows.

With this collaboration, HCLTech aims to deliver scalable, high-efficiency SoCs that reduce power consumption while maintaining performance across demanding workloads, India’s third-largest IT services company noted in a statement.

HCLTech added that it hopes to help enterprises respond to rising energy costs, sustainability requirements and the growing complexity of connected systems, including AI-driven workloads in data centres and edge computing environments.

“By partnering with HCLTech, we will be able to extend the reach of our low-power IP to more applications and customers than ever before,” said Pierre-Marie Dell’Accio, executive vice president of engineering at Dolphin Semiconductor.
“This partnership will help us push the boundaries of energy-efficient computing, whether it is for IoT devices or data centre ecosystems,” Dell’Accio added.

Hari Sadarahalli, corporate vice president and head of engineering and R&D services, HCLTech, noted, “As AI workloads surge, data grows exponentially, and sustainability becomes a top priority, our collaboration with Dolphin Semiconductor will empower our clients to lead with agility, high performance and a strong commitment to environmental responsibility.”

The companies did not disclose the financial terms of the partnership or any timeline for commercially launching the energy-efficient chips.

HCLTech has been reaching out to chip makers to augment its semiconductor and software design capabilities.

Earlier this year, it shook hands with semiconductor major AMD to develop solutions across AI, digital, and cloud.

Last year, the IT services company partnered with Arm, which offers high-performance and power-efficient intelligent processors, to leverage its custom silicon chips for supporting AI-driven business operations.

In 2024, it also collaborated with CAST, a semiconductor IP cores provider, to enable original equipment manufacturers to accelerate their automation journeys.

The post HCLTech, Dolphin Semiconductor to Co-develop Energy-Efficient Chips appeared first on Analytics India Magazine.

IBM and Riyadh Air have announced a partnership to introduce Riyadh Air, claiming to be the world’s first AI-native airline. Built entirely from the ground up without old patchwork systems, these AI-driven operations establish a new standard for enhancing guest and employee experiences while redefining innovation in the aviation sector.

Riyadh Air is leveraging IBM Consulting’s extensive industry knowledge and technical capabilities, a wide partner network, and IBM watsonx Orchestrate to function as an AI-native entity from the outset.

According to the announcement, initial flights are already in progress, and its first commercial service is anticipated in early 2026.

“By embedding AI into the very foundation of its operations, Riyadh Air is setting a new blueprint for what it means to build a modern, adaptive enterprise from the ground up,” said Mohamad Ali, senior vice president at IBM Consulting.

Riyadh Air is remoulding its travel experience by integrating generative and agentic AI into its operations to create seamless interactions between technology and human service. A personalised digital workplace will support employees with a chat-first platform for streamlined HR processes as the airline plans to double its workforce.

AI-powered mobile applications will enhance the employee and guest journey, using IBM watsonx Orchestrate to develop a proactive concierge experience that anticipates guest needs. Additionally, AI-driven voice bots and agent-assist tools will enable customer care agents to deliver personalised support, enhancing the traveller experience while upholding human values.

“We had a clear choice: be the last airline built on legacy technology or the first built on the platforms that will define the next decade of aviation,” said Adam Boukadida, chief financial officer at Riyadh Air.

According to the statement, the two companies have partnered to establish an AI-driven enterprise with the digital strategy and operational models to extend Saudi Arabia’s connectivity to over 100 destinations, serving millions of travellers by 2030.

The post Riyadh Air Partners with IBM to Bring Agentic AI to Commercial Flights by Early 2026 appeared first on Analytics India Magazine.

Explore Adobe Acrobat online PDF editor >

PDFs have long been the document standard for everything from resumes and government forms to business contracts and academic submissions. Their polished look and universal compatibility make them a go-to format. But when it comes to editing, things get tricky.

Editing a PDF often means dealing with unsecure third-party sites, running into compatibility issues, or fighting with tools that don’t really do what you need. This is where Adobe Acrobat’s free online PDF editor can help.

Adobe Acrobat offers an easy-to-use online PDF editor, right in your browser. It removes the barriers to PDF editing. It makes it possible for anyone to quickly add or remove text, annotate with comments or highlights, fill out forms, sign digitally, and more, without the need to install any software or sign up for a subscription.

Let’s explore why PDF editing has become essential, how Acrobat’s free online PDF editor makes it simple and accessible, and why upgrading to Adobe Acrobat Studio may be the right move for advanced needs.

Why Editing PDFs is More Important than Ever

Speed and flexibility make all the difference in today’s fast-paced workflows. From reviewing proposals to sending out critical reports and everything in between, the ability to edit PDFs on the go keeps projects moving forward. Seamless PDF editing, powered by advanced AI capabilities, supports smooth collaboration, reduces bottlenecks, and ensures all document needs keep pace with constantly evolving demands.

Everyday applications make this crystal clear. The Indian government is promoting efficiencies through paperless governance to simplify how documents are issued, stored, and shared through initiatives like DigiLocker, e-invoicing, and e-signing. As a result, more small businesses are moving their operations online and embracing digital workflows. In this digital-first environment, the ability to quickly edit documents is more important than ever.

Editing a PDF online saves time, cuts down on hassle, and helps you get polished results faster. No app installations, no device compatibility issues and no clunky workarounds. Just simple, seamless editing directly in your browser.

What is Adobe Acrobat’s Free Online PDF Editor?

Adobe Acrobat’s online PDF editor is an easy-to-use browser-based tool designed to make PDF editing simple, accessible, and secure.

Because it is entirely web-based, there is no need to install additional software. It works on both desktop and mobile, so you can open your PDF, make changes, and download the updated file wherever you are. The experience is smooth, consistent, and beginner friendly. With just a few clicks, you can open your file, make edits, and download your final version.

One thing that sets Adobe apart from others is the confidence that comes with using its PDF tools. Unlike unfamiliar apps or unverified solutions, Acrobat’s online editor is backed by Adobe’s long-standing reputation for document security and authenticity. You can edit with peace of mind knowing your files are handled with care and without the risks that often come with unreliable third-party software.

Key features of Adobe Acrobat’s Online PDF Editor

Even as a free tool, Adobe Acrobat’s online PDF editor comes packed with features. Here are some of them:

1. Give feedback.

Share your thoughts by adding sticky notes, adding text directly to the file, leaving suggestions, and tagging teammates to collaborate.

1. Annotate.

Insert text boxes anywhere in your PDF with custom fonts and formatting to suit your needs.

1. Highlight important sections.

Emphasise important text in multiple colours and attach notes to provide extra context.

1. Add markups.

Use freehand drawings, lines, or shapes in different colours and styles to bring attention to key points.

1. Fill and sign.

Complete digital forms by filling in fields, ticking boxes, and adding your name and signature.

1. Share with anyone.

Share a secure link to your PDF so anyone can view, comment, and collaborate without fuss.

Aside from these, Adobe has also made additional PDF editing tools available for free with no login required:

• Convert PDF to Word – Transform static PDFs into fully editable Word documents, eliminating the need for manual retyping and copy-pasting.
• Merge PDFs – Combine multiple PDFs into a single, organised document and save yourself from the hassle of juggling different files.
• Compress PDFs – Shrink large PDF files without losing quality or readability, making it easy to share documents via email or file-sharing portals.

How to use Adobe Acrobat’s Online PDF Editor.

Not sure how to edit a PDF file online? It is easy to use, even for beginners.

1. Go to Adobe Acrobat’s Online PDF Editor.
2. Choose a PDF to edit by clicking the “Select a file” button above or drag and drop a file into the drop zone.
3. Once Acrobat uploads the file, sign in to add your comments.
4. Use the toolbar to add text, sticky notes, highlights, drawings and more.
5. Download your annotated file or get a link to share it.

You will need to log in before you can start editing your PDF files.

Adobe Acrobat Studio: for tasks that go beyond the basics.

Adobe Acrobat’s free online PDF editor is great for quick edits, annotations, and sharing. But for small businesses looking to scale operations, streamline workflows, automate routine processes, access an exhaustive collection of design elements, or unlock AI-powered productivity, Adobe Acrobat Studio opens the door to an even more powerful set of features designed for professional and business use.

Here is why you might want to upgrade to Acrobat Studio:

• Unlock advanced AI capabilities.

Go beyond simple document manipulation. With Acrobat Studio, you can chat with your PDF for quick takeaways, summarize your PDFs for instant overviews, and create PDF spaces for an all-around knowledge and working hub.

• Strengthen safety and security.

Protect your work with enhanced security. Add or remove passwords, encrypt sensitive files, and control exactly who can view, copy, print, or edit your documents.

• Scale with your growing needs.

Acrobat Studio is designed to grow with you. It comes with the most comprehensive set of PDF tools, a prebuilt AI Assistant to help you reach your goals faster, and a design space powered by Adobe Express Premium for custom, high-quality images and templates. It has everything you need, whether you are jumpstarting your business or taking your operations to the next level.

Editing PDFs online: quick, easy, on the go.

With more of our work and everyday tasks happening online, it is safe to say that there is still a consistent need for effective PDF editing tools.

Adobe Acrobat’s online PDF editor proves that essential editing tools do not need to be complicated. You can highlight, comment, fill out forms, and sign documents right from your browser, without installing anything. It brings simplicity and flexibility straight to your browser, so that everyday document tasks can be done quickly and smoothly.

And when you are ready to go further, Adobe Acrobat Pro gives you the tools you need to work at a professional level.

Got questions? Check out the FAQs below:

How can I edit a date in a PDF file online?

To edit a date in a PDF, open the file using Adobe Acrobat’s free online PDF editor. Click on the text you want to change and simply type in the new date. Once done, save the file.

Can I edit PDFs on mobile for free?

You can edit PDFs right from your mobile browser using Adobe Acrobat’s online PDF editor, no downloads needed. Alternatively, you can view, comment on, and sign PDFs for free using the Adobe Acrobat Reader mobile app, available on both Android and iOS.

Is Adobe’s free online PDF editor secure?

Adobe encrypts your uploaded files and runs everything in a secure cloud environment. Whether you are editing a GST invoice, a job offer letter, or an Aadhaar verification form, Adobe helps your data stay private and protected.

Why is Adobe asking me to subscribe?

Some features, such as AI Assistant, are only available in Adobe Acrobat Studio. If you try to use one of these tools, Adobe may prompt you to upgrade. You can start with a free 7-day trial to explore all features.

The post Why Free PDF Editing with Adobe Acrobat’s Online PDF Editor is a Game Changer appeared first on Analytics India Magazine.

Google has introduced Gemini 3 Deep Think mode within the Gemini app, offering advanced reasoning capabilities for complex problem-solving. “Gemini 3 Deep Think is now available in the Gemini app,” wrote Tulsee Doshi, senior director of product management, in an announcement on December 4.

The new mode is available to Google AI Ultra subscribers and is designed to handle complex math, science, and logic queries. Users can access the feature by selecting “Deep Think” in the prompt bar and choosing Gemini 3 Pro from the model menu.

According to Google, Gemini 3 Deep Think demonstrates strong performance on benchmarks such as Humanity’s Last Exam, where it scored 41% without tools, and ARC-AGI-2, where it reached 45.1% with code execution.

The system uses parallel reasoning to test multiple hypotheses, building on earlier Deep Think variants that performed well in competitions like the International Mathematical Olympiad and the ICPC World Finals.

Google recently launched the broader Gemini 3 lineup, positioning it as a step toward more capable AI systems. The company said Gemini 3 Pro outperforms Gemini 2.5 Pro, OpenAI’s GPT-5.1, and Anthropic’s Claude Sonnet 4.5 across benchmarks, including LMArena, Humanity’s Last Exam, GPQA Diamond, and MathArena Apex.

With expanded multimodal inputs, extended context windows and new planning tools, Google said Gemini 3 can be applied to tasks such as analysing research papers, translating handwritten documents, generating visualisations, and evaluating sports performance.

The company also said Search’s AI Mode now includes generative UI features and interactive simulations.

The post Google Launches Gemini 3 Deep Think Mode for Ultra Subscribers appeared first on Analytics India Magazine.

The Telangana government announced on December 4 an extensive plan to transform Hyderabad into a ‘Quantum City’ and presented India’s long-term quantum strategy, aiming to establish the state as a key national and international hub for quantum technologies.

Speaking at the launch of NITI Aayog Roadmap for Quantum and the Telangana Quantum Strategy (TQS), the state’s IT minister, Sridhar Babu, said that quantum technology has the potential to be as revolutionary in this century as electricity and the internet were in previous ones.

He stated that the state government has adopted a targeted, long-term vision to ensure that Telangana becomes a global frontrunner in quantum technologies, AI, and cutting-edge digital systems.

“The NITI Aayog Quantum Roadmap is aligned with Telangana Rising 2047, the state’s vision to become a $3 trillion economy by 2047,” Mallu Bhatti Vikramarka, the deputy chief minister of Telangana, said, as reported by The New Indian Express.

The government will also establish a centre of excellence in quantum technologies as a national benchmark for research and skills development. Starting next financial year, a Fund of Funds will support deep-tech startups and emerging tech companies, as Babu announced.

Additionally, several media outlets reported that the state is establishing a Young India Startup Fund of Rs 1,000 crore, with a special focus on quantum and AI startups.

The strategy aims to enhance research infrastructure, cybersecurity, life sciences innovation, and human capital while promoting research in quantum computing, communication, sensing, and cryptography.

According to the state’s press release, the framework serves as a guiding roadmap for India’s engagement with quantum science and technology.

“With this launch, quantum technology will bring faster and deeper transformation across all sectors than any other technology and will also heavily influence national security, economic size and future growth,” Vikramarka added.

Similarly, the Karnataka government reportedly approved land in September in Hesaraghatta for a Quantum City (QCity) to establish Bengaluru as the “quantum capital of India.”

QCity will include advanced laboratories, quantum hardware production hubs, a high-performance computing data centre, and incubation facilities to support startups and foster collaboration between industry and academia.

The post Telangana Reveals Strategy Plans for Establishing Hyderabad as ‘Quantum City’ appeared first on Analytics India Magazine.

In a digital world dominated by text-based search and countless suggestions, our ability to visually comprehend information still lags behind. People can point their cameras at anything, yet rarely receive deeper meaning or cultural context in return.

Chance AI emerged this year from a similar human experience. As founder Xi Zeng recalled, standing before Sagrada Família in Spain 15 years ago left him awestruck, but online searches returned nothing more than “ticket links and souvenir advertisements,” a moment that convinced him the internet had become “great at selling, but very, very poor at explaining.”

His new venture, which already counts India as nearly a third of its user base, is built on that gap.

Chance AI positions itself as a visual reasoning engine designed not for transactions, but for context, culture and curiosity. Zeng describes it as “a combination of Google Lens and Instagram,” built around a simple loop: snap, understand and act.

While the company has secured seed funding, it is expecting to raise $7 to $10 million to fund its growth.

Building a Visual Intelligence Layer

Zeng believes visual reasoning, not text generation, will define the next shift in personal computing. “Our mission is to make AI that ensures curiosity, context and cultural understanding, not just convenience and productivity,” he said to AIM.

Chance AI uses a post-trained visual reasoning model developed in-house, while relying on major LLMs such as Gemini and GPT for the generative layer. The difference lies in the company’s ability to process data more efficiently than its competitors.

The company shared a benchmark comparison between major AI models when it comes to visual reasoning:

This foundation powers Chance AI’s growing ecosystem of “visual agents,” including popular features such as outfit feed tracking, skin assessments, and menu translation with dish imagery.

“All these agents are co-created with our community, and it’s like a small app store within the app,” Zeng said. So we are quite confident in the long-term competition because the things that we are building are not just built by us, it’s built by the whole community.

Competing With Big Tech by Moving Faster

Zeng is blunt about competing with large companies. “The only moat that we have is momentum,” he said, explaining how the company has no quick solution to stay ahead if big tech companies copy the idea.

He believes established firms are constrained by their business models. “Google makes profits from advertising. They have a specific business model that they can’t escape from.”
Chance AI, instead, focuses on curiosity-driven search, visual memory summaries, and a community layer built around visual taste, elements he says Google would never pursue.

Future Outlook

Zeng’s relationship with India began during his years at OnePlus and TikTok. He described the country not merely as a user base, but as a creative force. “India is not just a market, it’s a movement,” he said.

India now sits at the centre of Chance AI’s creator strategy, with the company accelerating its on-ground momentum through student-led communities, multi-platform expansion and a focused push to solve creators’ biggest constraint, the lack of time. With more than 100 million active creators in the country, the platform aims to help them turn ideas into publish-ready visuals within seconds, removing the friction of continuous ideation, shooting and editing.

Its upcoming “Chanced by Creators” program will onboard leading Indian creators with early access, challenges and incentives, culminating in a Times Square UGC showcase spotlighting global submissions. The initiative reflects a broader ambition to position Indian creator talent on a world stage and reinforce a simple message, “Don’t think. Just Chance it.”

In that context, Chance AI is now partnering with Indian design universities and communities to develop what Zeng calls a co-creation strategy, aiming to absorb “India’s sense of design, rhythm and storytelling.”

The company claims to have approximately 200,000 users and aims to reach one million. Hardware is the next frontier, with Zeng envisioning a wearable “third eye” that understands the world in real time. He even sees Chance AI becoming a key layer atop devices like Meta’s Ray-Ban glasses.

Zeng wants to focus on growth first and then look at monetisation. He’s betting on Meta’s growth to chart his company’s success. Chance AI’s longer-term ambition is to license its visual reasoning stack to future AI hardware, including glasses from mainstream brands. The team is also exploring open-sourcing parts of its technology to support visually impaired users.

The post This Startup is Trying to Build a Visual Version of Perplexity appeared first on Analytics India Magazine.

AWS has introduced its fifth-generation Graviton processor, Graviton5, which the company says delivers up to 25% higher performance than its previous generation while improving energy efficiency.

The chip will power new Amazon EC2 M9g instances, now available in preview. C9g (compute-focused) and R9g (memory-focused) instances are planned for 2026.

AWS said the launch comes as organisations look for faster performance and lower costs at scale. “Graviton5 delivers up to 25% better compute performance than the previous generation while maintaining energy efficiency,” the company said in a blog post.

Core Specs and Performance Gains

The new processor includes 192 cores, offers a 5x larger L3 cache, and provides faster memory speeds. AWS said the design reduces inter-core communication latency by up to 33%, enabling workloads such as gaming, big data analytics, databases, and EDA tools to scale with higher throughput.

Network bandwidth increases by up to 15% on average, and Amazon EBS bandwidth increases by up to 20%. For the largest instances, network bandwidth doubles.

AWS said the chip is built on 3nm technology, and its server architecture uses bare-die cooling to improve efficiency.

Graviton5 instances run on the AWS Nitro System and include a new Nitro Isolation Engine, which uses formal verification to mathematically ensure workload isolation. AWS said this provides “a new standard for mathematically proven cloud security.”

Early Customer Results

AWS said more than half of its new CPU capacity added in the past three years is powered by Graviton. According to the company, 98% of the top 1,000 EC2 customers, including Adobe, Epic Games, Formula 1, Pinterest, Snowflake, and Siemens, already use Graviton-based instances.

Airbnb reported performance improvements of up to 25% during tests using its production search workloads.

Atlassian said Jira testing on M9g instances showed 30% higher performance and 20% lower latency than the previous generation. “We look forward to AWS Graviton5 general availability,” said Paulo Almeida, principal site reliability engineer.

SAP said it saw OLTP query performance 35% to 60% better on SAP HANA Cloud. Siemens Digital Industries Software said early Graviton5 tests delivered another 30% performance boost for its Calibre platform.

Synopsys reported up to 35% faster runtimes for EDA workloads and said Arm observed up to 40% faster runtimes for Synopsys VCS.

The post AWS Launches Graviton5, Its Most Powerful Custom CPU for EC2 appeared first on Analytics India Magazine.

AWS used re:Invent 2025 to signal that the next phase of its AI strategy will be built on speed, scale and a new layer of agent-based computing.

The company launched its Trainium3 UltraServers with huge jumps in performance and energy gains, shared first details of Trainium4, expanded Bedrock into the largest neutral model hub, and pushed deeper into agent tooling across policy, evaluation and autonomous workflows.

AWS said it has already deployed more than 1 million Trainium chips, and the new stack is meant to cut costs, reduce latency, and support systems that run on massive parallel agents.

Matt Garman’s keynote outlined where AWS sees the world going and what it wants to build for it.

A Deeper Push into NVIDIA and Large Scale AI Training

Garman opened by framing the NVIDIA partnership as central to AWS. He said AWS and NVIDIA have worked together for more than 15 years and that nothing about the collaboration is accidental.

“Nothing’s too small for us to really work together to make sure that we have the most reliable performance.” He added that NVIDIA itself trains its largest systems on AWS, calling it “a testament to work together.”

The message was that AWS wants to be the most stable home for frontier model training, with NVIDIA hardware blending into AWS silicon and networking.

AI Factories for Customers that Want Hyperscale Training Inside Their Own Walls

Garman said many governments and large enterprises have the data centre footprint but lack the know-how to run giant AI clusters. He said the idea came from working with players like OpenAI and Humain, the Saudi AI initiative.

“Why can’t we help more customers, the ones who really need this large-scale infrastructure, see what our expertise, our services, are understanding?” he said.

AI factories let AWS place its infrastructure, software and governance controls inside a customer environment while meeting rules on sovereignty and policy. AWS wants to make hyperscale AI feel like owned infrastructure rather than a remote service.

The Next Generation of AWS Silicon with Trainium3 and Trainium4

Garman previewed Trainium4 while Trainium3 went live. He said Trainium4 delivers “over 6x the FP, 4x performance, 4x more memory family and 2x more high bandwidth memory capacity” and doubles power efficiency compared to the earlier generation.

He also showed how fast inference now resembles training loads. “There’s not going to be an experienced application of a system built that doesn’t rely on inference.” AWS is positioning its chips as the backbone of low-cost training, low-latency inference and huge agent workloads.

Bedrock Becoming the World’s Largest Mix and Match Model Platform

Bedrock now has more than 100,000 customers. Garman said it has doubled the number of models in a year and will add 18 more new open weight models, including Google, MiniMax, Mistral, NVIDIA and OpenAI gpt-oss.

He said customers are increasingly running many models at once. “This mix and match is going to be normal.” AWS also refreshed its own Nova family. Nova Light is for cost-efficient reasoning. Nova Pro targets complex reasoning across documents and video. Nova Sonic adds multilingual speech-to-speech.

The unified Nova multimodal model handles text, images, video and speech as inputs. Garman said this solves a real need for creative teams who want one model that “can output different forms of text and imagery” without juggling multiple systems.

Nova Forge, a New Way for Companies to Build Their Own Frontier Model

This was one of Garman’s biggest claims. Customers want their models to reflect their own language and systems, but fine-tuning breaks when pushed too far. Garman said the team asked a simple question. “Why not make that possible? Why can’t that be true?”

Nova Forge gives access to Nova checkpoints and lets customers blend their own data with Amazon-curated sets, then deploy the resulting frontier model on Bedrock with full guardrails. This lets enterprises create models that act like internal experts rather than generic assistants.

The Full Stack for Building, Governing and Monitoring Billions of Agents

Garman said the world is entering a time “where there were literally billions of agents working together.” AWS wants to make those agents safe, fast and easy to build. Bedrock Agent Core brings building blocks like memory, gateway and identity. New upgrades include policy and evaluations.

With policy, he said, customers can set rules in simple language. “We do the hard work to translate it into policy code.” Evaluations monitor an agent’s behaviour in the real world and raise alerts when quality slips. “

You’re going to get an alert that says the agent review isn’t acting as it should.” AWS sees this as the missing layer for running agent systems at scale.

Frontier Agents, AWS’s Next Step in Autonomous Software

Garman said the company learned from internal use of Kiro that teams were still treating agents like simple assistants. AWS then changed the design. Agents should be autonomous, handle long tasks, work across hundreds of parallel actions, and improve without human babysitting.

“I don’t have to overwork, I don’t have to babysit.” The result is Frontier agents.

The Kiro autonomous agent keeps persistent context, pulls requests, improves code and learns how a team works. The security agent embeds a security expert in every step of development and can run pen tests on demand. The DevOps agent handles incident triage and recovery.

Garman said it offers “fewer alerts” and faster recovery across multi-cloud and hybrid setups. AWS wants these agents to give step-change productivity, not small gains.

The post 7 Things Matt Garman Announced AWS Is Focusing On appeared first on Analytics India Magazine.