LTIMindtree, Armada To Scale Edge & Sovereign AI Deployments

Edge infrastructure company Armada has partnered with LTIMindtree to accelerate global adoption of edge AI, sovereign AI, and federated learning, aiming to help large enterprises run advanced AI workloads in remote and regulated environments.

Under the collaboration, Armada will integrate its Galleon modular data centres and Armada Edge Platform (AEP) with LTIMindtree’s global delivery network and digital transformation services.
In a release LTIMindtree chief growth officer Krishnan Iyer said that by working with Armada, “we can help enterprises unlock the value of Edge AI across global and distributed operations.”

LTIMindtree, which serves a broad portfolio of enterprise clients, will bring Armada’s edge products to existing customers while supporting expansions into markets where Armada does not yet operate.

Armada’s AEP combines compute, connectivity, and real-world AI into a unified system designed for rugged, distributed, and intermittently connected environments, common in manufacturing, healthcare, energy, and other regulated sectors.

The companies said the joint offering is intended to help enterprises maintain data sovereignty, privacy, and operational continuity while deploying AI at scale.

“Our work with LTIMindtree strengthens Armada’s ability to help enterprises accelerate AI adoption at the edge,” Pradeep Nair, founding CTO of Armada, said. “Combining Armada Edge Platform with LTIMindtree’s delivery and integration capabilities will enable more organisations to deploy real-world AI with full governance over their data and infrastructure.”

Both companies said the collaboration is aimed at capturing rising demand from enterprises looking to run AI close to the source of data, especially in markets with strict compliance requirements and unreliable connectivity.

The post LTIMindtree, Armada To Scale Edge & Sovereign AI Deployments appeared first on Analytics India Magazine.

NVIDIA Open Sources Reasoning Model for Autonomous Driving at NeurIPS 2025

NVIDIA OpenUSDNVIDIA OpenUSD

At NeurIPS 2025, NVIDIA announced a new set of open models, datasets and tools spanning autonomous driving, speech AI and safety research, strengthening its position in open digital and physical AI development.

The company also received recognition from Artificial Analysis’ new Openness Index, which placed NVIDIA’s Nemotron family among the most transparent model ecosystems.

NVIDIA released DRIVE Alpamayo-R1, described by the company as “the world’s first open reasoning VLA model for autonomous driving.”

Bryan Catanzaro, NVIDIA’s vice president of applied deep learning research, said the model integrates chain-of-thought reasoning with path planning to support research on complex road scenarios and level-4 autonomy.

According to NVIDIA, AR1 breaks down scenes step by step, considers possible trajectories and uses contextual data to determine routes. A subset of its training data is available through NVIDIA’s Physical AI Open Datasets, and the model will be accessible on GitHub and Hugging Face.

Built on NVIDIA Cosmos Reason, AR1 can be customised for non-commercial research. NVIDIA said reinforcement learning was effective in post-training the model, improving its reasoning performance compared with the pretrained version. The company also released AlpaSim, an open framework for evaluating AR1.

Moreover, NVIDIA expanded the Cosmos ecosystem with new tools and workflows in the Cosmos Cookbook, offering step-by-step guidance for model post-training, synthetic data generation and evaluation.

New Cosmos-based systems include LidarGen, a world model for generating lidar data; Omniverse NuRec Fixer, for correcting artifacts in neural reconstructions; Cosmos Policy for turning video models into robot policies; and ProtoMotions3, a framework for training physically simulated digital humans and robots.

Industry partners, including Voxel51, 1X, Figure AI, Foretellix, Gatik, Oxa, PlusAI and X-Humanoid, are using Cosmos world foundation models. ETH Zurich researchers are presenting NeurIPS work showing how Cosmos models can generate cohesive 3D scenes.

In digital AI, NVIDIA introduced new models and datasets under the Nemotron and NeMo umbrellas. These include MultiTalker Parakeet, a speech recognition model for multi-speaker environments; Sortformer, a diarization model; and Nemotron Content Safety Reasoning, which the company said applies domain-specific safety rules using reasoning.

NVIDIA also opened the Nemotron Content Safety Audio Dataset, used for detecting unsafe audio content. Tools for synthetic data and reinforcement learning were also released, including NeMo Gym for RL environments and the NeMo Data Designer Library, now open-sourced under Apache 2.0.

CrowdStrike, Palantir and ServiceNow are among partners using Nemotron and NeMo tools for specialised agentic AI.

NVIDIA researchers are presenting more than 70 papers and sessions at NeurIPS.

The post NVIDIA Open Sources Reasoning Model for Autonomous Driving at NeurIPS 2025 appeared first on Analytics India Magazine.

HSBC, Mistral AI Sign Multi-Year Deal to Scale AI Across Global Operations

PartnershipPartnership

HSBC and French AI start-up Mistral AI have entered a multi-year strategic partnership aimed at accelerating the adoption of generative AI across the bank’s global operations.

The tie-up gives HSBC access to Mistral’s commercial AI models, including future iterations, and establishes joint development pipelines between the two companies’ applied AI, science and engineering teams.

The partnership will allow HSBC to run self-hosted, enterprise-grade models on its own internal infrastructure, a move the bank says is critical for privacy, governance and scale. HSBC has spent the past year assessing a broad set of LLMs as part of its technology roadmap.

According to the company, Mistral’s expertise in foundational model development offered a path to strengthen its internal AI tools, including an AI productivity platform now used by employees worldwide.

The collaboration aims to streamline business processes and automate high-volume, document-heavy workflows. HSBC said its teams will use Mistral’s models to generate customised business tasks for client-facing, procurement and marketing functions; enhance financial analysis in complex lending workflows; and improve multilingual reasoning and translation capabilities for customer-support teams handling global interactions.

The bank also expects gains in innovation cycles, enabling faster prototyping and deployment of new features across units.

Future phases of the partnership will explore customer-facing applications, including improvements to credit and lending processes, onboarding journeys, and more robust fraud and anti-money-laundering checks.

Georges Elhedery, HSBC Group CEO, called the partnership “an exciting step forward in HSBC’s technology strategy,” adding that the collaboration would “equip our colleagues with tools to innovate, simplify daily tasks, and free up time to deliver for our customers.”

Mistral AI CEO and co-founder Arthur Mensch said the company’s “highly customisable, enterprise-grade frontier AI solutions will reinvent HSBC’s workflows and services while ensuring full ownership of data.”

Both organisations underscored their commitment to responsible AI deployment, including transparency, privacy and strong governance.

The post HSBC, Mistral AI Sign Multi-Year Deal to Scale AI Across Global Operations appeared first on Analytics India Magazine.

Indian E-Commerce Firm Meesho Expands Vernacular AI Voice Agents Ahead of IPO

Indian e-commerce firm Meesho has developed a real-time voice agent using ElevenLabs text-to-speech to automate customer support in Hindi and English.

It efficiently manages many inquiries without human intervention, covering order delays, cancellations, and refunds, while reducing average handling time, according to Eleven Labs.

“Voice AI is changing how businesses in India interact with their customers. Meesho’s deployment shows what’s possible when multilingual, natural-sounding speech becomes the interface,” said Siddharth Srinivasan, GTM India, ElevenLabs.

The company is set to open subscriptions for its three-day initial public offering (IPO) sale starting on December 3, with a price band of ₹105 to ₹111 per equity share.

Ahead of the IPO, the company’s report said it is deepening its AI capabilities through its Meesho AI Labs initiative, aiming to institutionalise long-term, frontier AI innovation.

The team has built AI-driven systems that personalise the shopping experience, such as optimising product rankings and tailoring recommendations.

Meesho AI Labs has also worked to improve the customer experience through vernacular voicebots that leverage generative AI to resolve queries more naturally and intuitively for shoppers in smaller towns and rural areas.

Key focuses of Meesho AI Labs also include creating language models tailored for Indian consumers, enhancing the shopping journey with agentic AI, improving transaction risk management, and increasing the effectiveness of advertising through AI.

The company intends to build and roll out technology products, including various logistics management systems and content creator technologies, to optimise fulfilment, content generation, and seller operations at scale, Meesho said in a report.

“We are also currently piloting the externalisation of our AI-enabled support services to third parties,” the company added.

Additionally, Meesho Networks was set up in April 2025, “with the objective of commercialising AI technologies and solutions.”

The company intends to continue building its technology stack to “reduce manual overheads and improve process efficiency across the platform.”

They aim to develop technology products, including logistics management systems and content creator tools, to enhance fulfilment, content generation, and seller operations.

Designed for Meesho’s low-cost environment, these systems will leverage automation to expedite cataloguing and product discovery, the report said.

As per Reuters, Meesho aims for a valuation of up to $5.6 billion in its upcoming IPO, which is projected to reach $170-$190 billion by 2030, according to a Bain and Flipkart report.

The post Indian E-Commerce Firm Meesho Expands Vernacular AI Voice Agents Ahead of IPO appeared first on Analytics India Magazine.

Accenture Partners with OpenAI to Roll Out ChatGPT Enterprise for Its Workforce

Accenture and OpenAI have formed a wide collaboration that aims to push enterprise AI deeper into large companies, starting with Accenture’s own workforce.

The firm will roll out ChatGPT Enterprise to tens of thousands of its employees, which the company says is the largest group to be upskilled through OpenAI Certifications. Accenture will use the tool across consulting, operations and delivery work as it builds new AI services for clients.

Accenture has named OpenAI as one of its primary partners for its next wave of AI-powered offerings. The company says the partnership will help it bring agentic AI into the core of business operations and speed up adoption for large enterprises.

Julie Sweet, chair and CEO of Accenture, said, “By combining OpenAI’s breakthrough technologies with Accenture’s deep industry and functional expertise and global delivery capabilities, we will accelerate enterprise reinvention and business outcomes for our clients.”

Fidji Simo, CEO of Applications at OpenAI, said, “Accenture plays an important role in helping companies adopt the technologies that define each new era, and we’re excited to partner with them to accelerate the AI transformation of the largest enterprises.”

The companies are launching a flagship AI program that blends OpenAI’s enterprise products with Accenture’s AI experience. Accenture will get access to OpenAI implementation playbooks, industry use-cases, deployment guidance and hands-on expertise so it can bring AI into real workflows for clients.

Both firms will also work on new solutions for customer service, supply chain, finance, HR and other key business functions. Accenture will use OpenAI’s AgentKit to help clients design and deploy custom AI agents that automate work and support decision making.

According to reports, Accenture has also started rebranding its 800,000 staff and calling them ‘reinventors’ to adapt with AI.

The companies say the goal is to help joint clients adopt OpenAI’s agentic capabilities faster and integrate them across their operations. OpenAI already works with some of the world’s largest enterprises, including Walmart, Salesforce, PayPal, Intuit, Target, Thermo Fisher, BNY, Morgan Stanley and BBVA.

The post Accenture Partners with OpenAI to Roll Out ChatGPT Enterprise for Its Workforce appeared first on Analytics India Magazine.

Legal Heat on TCS, Wipro Prompts Wider IP Rethink in Indian IT

Indian IT’s confidence in its delivery discipline is being tested as TCS faces a major setback in a trade secrets lawsuit, and a fresh patent suit in the US, while Wipro has been named in a new patent-infringement case overseas.

The developments are prompting the industry to reassess whether its governance, engineering culture and compliance frameworks can keep pace with the far stricter intellectual-property expectations of global clients, especially as firms move deeper into platforms, products and proprietary AI systems where IP boundaries are tighter and scrutiny is sharper.

Technology leaders, engineering heads and legal experts say the sector has matured, but gaps remain, particularly, as companies move deeper into platform and product development, where IP boundaries are far tighter than in traditional services.

Ritesh Gupta, CTO of Happiest Minds Technologies, said Indian IT providers have grown significantly more mature in safeguarding client IP as they build products and platforms, driven by stricter client expectations, tighter contracts and heightened IP sensitivity in global technology deals. Happiest Minds Technologies defines itself as “a mindful IT company,” that enables digital transformation for enterprises and technology providers.

According to Gupta, tier-1 firms already operate with strong contractual frameworks, secure development processes and clear IP-ownership governance, and mid-sized firms are rapidly catching up.

At Happiest Minds, he said the company follows rigorous legal vetting of client contracts, secured and isolated development environments, role-based access, regular audits and strict mechanisms for open-source scanning and SBOM management.

Gupta said that engineers receive specific training on IP protection, and that the firm restricts public generative AI tools, while maintaining meticulous code tracing to avoid contamination.

He acknowledged that the industry still struggles with challenges such as unintentional code reuse, weak open-source license governance and risks introduced by generative AI. The services mindset is evolving with the discipline required for product development, he noted.

Gupta said the company’s approach embeds compliance directly into engineering processes, so speed never compromises client IP. He argued that the entire industry needs stronger governance, automated compliance tooling, better awareness and a more IP-conscious culture to prevent violations.

Debashis Singh, chief information officer at Persistent Systems, said the industry’s shift from traditional IT to modern platform engineering has demanded a new level of rigour, as legacy architectures built for stability offer little flexibility and often carry institutional knowledge known only to a few.

He observed that the cloud era introduced open, modular architectures and automation that radically compressed delivery timelines, placing modern engineering at the intersection of speed and discipline.

Singh said Persistent maintains strict access limits and clear IP boundaries through zero-trust frameworks, secure private environments and enterprise-grade AI models that ensure no customer environment is used to train AI systems.

He acknowledged that teams historically struggled to understand undocumented legacy systems, but argued that generative AI and intelligent platforms, such as Persistent’s SASVA, now accelerate reverse engineering and documentation, transforming what was once a slow, opaque process.

According to him, 70–80% of legacy understanding can now be automated, though the final 20–30% still requires human expertise.

He cautioned that reliance on reverse engineering introduces risks such as misinterpreted logic or inadvertent IP exposure, which Persistent mitigates through structured governance, DevOps-led automation and strong knowledge management.

Singh said adherence to IP rules ultimately depends on user awareness, ethical AI guardrails, continuous improvement and embedding compliance into everyday engineering, enabling delivery at speed without compromising legal boundaries.

Priyanka S Kulkarni, a senior legal advisor specialising in IP, said the Indian IT industry broadly recognises the legal risk of exceeding authorised access, especially under US and international regulations. But, it must remain vigilant in documenting permissions, monitoring access and ensuring employees understand the exact boundaries of their authorisations, she added.

She noted that while most vendor–client contracts contain confidentiality clauses, many lack precise definitions of permissible access during modernisation or integration work, creating potential exposure around trade secrets and IP ownership.

According to her, blind spots commonly include overly broad internal access, unclear ownership of accelerators and derivative works, weak exit protocols, and scenarios where multiple vendors share overlapping environments.

She said firms should rely on strict role-based access controls, immutable logs, automated alerts, periodic audits and clear revocation processes to create defensible evidence of compliance.

As companies shift from services to platforms and productised offerings, Kulkarni argued that they must adopt more structured IP frameworks covering background and foreground IP, residual rights, co-developed assets and licensing boundaries, backed by robust data-governance, liability and indemnity clauses. She said the platform era demands far tighter legal, contractual and compliance scaffolding than the traditional services model.

The post Legal Heat on TCS, Wipro Prompts Wider IP Rethink in Indian IT appeared first on Analytics India Magazine.

TwelveLabs Launches Marengo 3.0 Video Understanding Model on TwelveLabs and Amazon Bedrock

Fujitsu Develops Multi-AI Agent Collaboration Tech to Optimize Supply Chains, Launches Joint Trials

KAWASAKI, Japan, Dec. 1, 2025 — Fujitsu Limited today announced the development of a multi-AI…

LTTS Unveils NVIDIA-Powered Digital Twin Platform for Advanced Respiratory Diagnostics

L&T Technology Services (LTTS) has announced the development of a next-generation AI-powered digital twin platform for respiratory diagnostics and lung navigation, developed in collaboration with NVIDIA.

The company said the platform aims to deliver scalable, low-latency solutions that improve diagnostic precision and expand access to advanced imaging tools for clinicians worldwide.

The solution, to be showcased at the Radiological Society of North America (RSNA) 2025 conference, integrates with CT imaging and uses deep learning models to generate a 3D digital twin of lung anatomy.

Built on NVIDIA MONAI for medical image segmentation and NVIDIA TensorRT for optimised AI inference, the platform enables detailed visualisation of airways, blood vessels, lobes and lesions, along with interactive simulation and path-planning support for bronchoscopy procedures.

LTTS said its engineering capabilities in medical imaging and proprietary navigation systems allow static CT scans to be converted into dynamic, clinically meaningful digital representations that evolve with patient data.

The company said such biological digital twins can support planning and navigation for conditions including lung cancer, COPD and infectious diseases.

“AI is reshaping what’s possible in diagnostics and medical technology,” said Alind Saxena, executive director and president of mobility and tech at LTTS.

“Our collaboration with NVIDIA combines LTTS’ expertise in AI-driven diagnostics and predictive analytics, with NVIDIA’s powerful modeling and visualisation platform. This lets us engineer a digital twin platform that not only enhances diagnostic accuracy but also gives clinicians an immersive, real-time planning tool ultimately helping deliver better outcomes for patients around the world,” Saxena added.

David Niewolny, director of business development for healthcare/medical at NVIDIA, said the partnership demonstrates the potential for accelerated computing in clinical workflows.

“Working with LTTS to accelerate the development of AI-enabled medical technology demonstrates how NVIDIA is empowering the healthcare industry with accelerated computing and AI innovation,” Niewolny said.

“LTTS is enabling transformative solutions that bring the vision of real-time AI and biological digital twins powered by NVIDIA to clinical practice, delivering interactive, real-time visualisation and intelligent guidance to help clinicians provide higher-quality care and achieve better outcomes for patients.”

The post LTTS Unveils NVIDIA-Powered Digital Twin Platform for Advanced Respiratory Diagnostics appeared first on Analytics India Magazine.

DeepSeek Releases New Reasoning Models to Match GPT-5, Rival Gemini 3 Pro

Chinese AI lab DeepSeek has launched two new reasoning-first AI models, DeepSeek-V3.2 and DeepSeek-V3.2-Speciale, expanding its suite of systems for agents, tool-use and complex inference.

Both the models and the accompanying technical report have been released as open source on Hugging Face.

The company announced on X that V3.2 is the official successor to V3.2-Exp and is now available across its app, web interface and API. The Speciale variant is offered only through a temporary API endpoint until December 15, 2025.

DeepSeek said V3.2 aims to balance inference efficiency with long-context performance, calling it “your daily driver at GPT-5 level performance.”

The V3.2-Speciale model, positioned for high-end reasoning tasks, “rivals Gemini-3.0-Pro,” the company said. According to DeepSeek, Speciale delivers gold-level (expert human proficiency) results across competitive benchmarks such as the IMO, CMO and ICPC World Finals.

The models introduce an expansion of DeepSeek’s agent-training approach, supported by a new synthetic dataset spanning more than 1,800 environments and 85,000 complex instructions. The company stated that V3.2 is its first model to integrate thinking directly into tool use, allowing structured reasoning to operate both within and alongside external tools.

Alongside the release, DeepSeek updated its API, noting that V3.2 maintains the same usage pattern as its predecessor. The Speciale model is priced the same as V3.2 but does not support tool calls. The company also highlighted a new capability in V3.2 described as “Thinking in Tool-Use,” with additional details provided in its developer documentation.

The company recently also released a new open-weight model, DeepSeekMath-V2. The model, as per the AI lab, demonstrates strong theorem-proving capabilities in mathematics and achieved gold-level scores on the International Mathematics Olympiad (IMO) 2025.

The post DeepSeek Releases New Reasoning Models to Match GPT-5, Rival Gemini 3 Pro appeared first on Analytics India Magazine.