From News Desk
NimbleEdge, the on-device AI platform built for performance, privacy and portability has announced the global launch of its on-device agentic AI platform ‘DeliteAI’. It is an open-source infrastructure that enables developers to build, deploy and run fully customised AI-native experiences directly on smartphones without relying on the Cloud.
With the launch of DeliteAI, NimbleEdge is putting a full-stack, production-ready on-device AI platform into the hands of developers and ML engineers. The open source stack enables fast, cross-platform deployment of transformer models, LLMs and multimodal AI, all without relying on cloud infrastructure or high-end GPUs.
This comprehensive launch includes three major components –
- A production-ready SDK with an optimised inference stack and the industry’s first on-device Python runtime for orchestrating agentic workflows
- A dedicated Agent Marketplace where developers can discover and integrate pre-built AI agents in their mobile applications
- The NimbleEdge Assistant, the world’s first fully on-device conversational AI assistant with built-in productivity capabilities
“We believe AI shouldn’t sit miles away from us in data centres ,” said Varun Khare, Founder and CEO, NimbleEdge. “With this launch, any mobile application can scale AI to billions of users while improving data safety and user privacy. They own the entire stack, from the models they use to the way intelligence shows up in their products bringing a unique AI enabled experience to their users.”
Designed to address the limitations of Cloud-based AI, which requires constant connectivity, increases latency, creates privacy risks and incurs unsustainable operational costs, NimbleEdge also fills critical gaps in the existing on-device AI ecosystem. Until now, developers lacked unified tooling, standardised runtimes; and a robust marketplace of ready-to-integrate agents to build sophisticated AI-native experiences on smartphones.
NimbleEdge’s platform empowers any company or developer to bring their own models, including Llama, Gemma, or Qwen and run all inference directly on the user’s device, abstracting away the complexities of diverse mobile hardware and managing runtimes like ONNX, LiteRT, or ExecuTorch. This architecture ensures that no personal data ever leaves local hardware, allowing organizations to fine-tune and deploy large language models and agents entirely offline. Unlike proprietary assistants tied to a single ecosystem, NimbleEdge makes it possible to create completely customized workflows and branded assistants all powered by an open-source, on-device ecosystem without external dependencies.
Neeraj Poddar, Co-Founder and CTO at NimbleEdge, added, “For the first time, developers can bring state-of-the-art AI models to consumer devices, orchestrate them with Python and deploy truly private AI agents at scale. This is the missing infrastructure layer and developer tooling we wished existed when we were building distributed systems at global scale, and now it’s open for everyone.”
NimbleEdge has already demonstrated this capability at scale, powering AI & ML experiences across more than 30 million devices in production deployments for gaming and e-commerce apps, with AI infrastructure partners such as PyTorch and ONNX. The platform’s on-device Python runtime allows developers to build dynamic, real-time applications with familiar tools while maintaining full control over data flows and the Agent Marketplace provides a growing library of plug-and-play agents for tasks like summarisation, recommendations, and speech processing.
Aakrit Vaish, early investor in NimbleEdge and a member of the India AI Mission commented, “NimbleEdge reflects the innovative work happening in India’s AI ecosystem. By bringing agentic AI capabilities on-device, they’re addressing important challenges around privacy and latency and creating open source infra that can reach users at scale in India and globally.”
Global Perspectives... 