Featured publication

Expanding The Horizons of Generative Edge AI:Mission, Vision, and Insights From Industries Journal of Systems Engineering and Electronics (ISSN NO: 1671-1793) Volume 35 ISSUE 8 2025 1-JSEE3316_Download

Generative Edge AI Forum #4 – Architectures, Agents & Apps

November 2025 https://youtube.com/playlist?list=PLeisuBi-nfBNiPURmxFr4dp7qwMc3aaZz&si=8ugkQyIvXK0MwSc2 The Edge AI Revolution is Here The cloud’s dominance is being challenged—and the Generative Edge AI community is leading the charge. Across three groundbreaking EDGE AI FOUNDATION forums, we’ve witnessed a seismic shift: the cloud must evolve beyond its role as a centralized AI powerhouse. After years

Heterogeneous Low-Power Agentic Architecture for Edge Distributed Home Surveillance

Abstract Recent advancements in edge hardware enabled increasingly complex artificial intelligence workloads to be executed directly on resource-constrained devices. However, single-edge devices remain computationally limited compared with hybrid or cloud alternatives. This work presents a fully edge-based architecture implementing agentic AI on heterogeneous devices for home surveillance with a natural

Recursive Multi-Agent Systems

RecursiveMAS, a recursive multi-agent framework that casts theentire system as a unified latent-space recursive computation 2604.25917v2Download

Beyond Edge AI? Reflections on Toward Edge General Intelligence with Agentic AI and Agentification

Every now and then, a survey paper appears that does more than summarize a research field—it attempts to define where that field is heading. The recently released survey Toward Edge General Intelligence with Agentic AI and Agentification: Concepts, Technologies, and Future Directions is one of those papers. Spanning more than

Google Coral Is Back. Why This Matters for Embedded AI.

For many of us working in Edge AI, the name Coral brings back memories of one of the first platforms that made hardware-accelerated on-device machine learning genuinely accessible. The USB Accelerator, the Dev Board, and later the Dev Board Micro became reference platforms for countless prototypes, demos, and research projects.

Small AI Models Gain Traction Around the World 

In places with unreliable networks and no data-center infrastructure, smaller is better Small Language Models Power Life-Saving Small AI - IEEE SpectrumDownload

GENERATIVE AND AGENTIC EDGE AI WORKSHOPS – JULY 23rd

REGISTER FOR FREE HERE: https://streamyard.com/watch/tzSrJr8RJAeJ Join us for three back-to-back hands on workshops with Intel, STMicroelectronics, MiTO and ForestHub to learn the latest techniques and technologies in generative and agentic edge AI. Please register but note that these workshops are FIRST COME FIRST SERVED based on availability of virtualized platforms.

LFM2.5-230M Physical AI

LFM2.5-230M https://www.liquid.ai/blog/lfm2-5-230m on a Unitree G1 https://www.youtube.com/shorts/CuMOWa2y1Ho, running on NVIDIA Jetson Orin. it takes a single natural-language instruction and decomposes it into a sequence of tool calls that invoke pre-trained low-level skills provided by NVIDIA's SONIC framework LFM2.5-230M_ Built to Run Anywhere _ Liquid AIDownload

Google’s Training Supercomputers from TPU v2 to Ironwood

2606.15870v1Download

New DiffusionGemma and MoQ GGUFs for Gemma 4 12B and LFM2.5 8B A1B

DiffusionGemma and MoQDownload

IEEE Spectrum 6/2026

06_Spectrum_26-medDownload

Complexity Analysis for Categorized Edge Language Models

Niks Kordjukovs and Danilo Pietro Pau symmetry-18-00766 (5)Download

The State of AI in 2026

Source JUNE 2026  SPECTRUM.IEEE.ORG  pag 11 Screenshot 2026-06-01 182622Download

AI Chips at the Edge of Success

IDTechEx_WebinarAIChipsattheEdgeofSuccess_PDFDownload

Scaling Down is the New Scaling Up

by Michael Stauffer mkstauffer@alum.mit.edu T1T02_Chandra_Meta_2026Download

A Useful Compass for On-Device GenAI: Reviewing Awesome Mobile LLMs

Over the past couple of years, we have seen an explosion of work around running Large Language Models beyond the cloud. What used to be a rather clear separation, i.e., heavy models in the cloud and lightweight inference at the edge, is now increasingly blurred. In this context, I recently

Vision Language to Actions: A practical guide (4/22/26 Forum 5 Gen Edge AI by Danilo Pau)

PAU_VLA_Forum_5Download

Mercedes-Benz and Liquid AI have entered a multi-year strategic partnership as of April 2026.

The collaboration focuses on integrating Liquid AI’s Liquid Foundation Models (LFMs) into Mercedes-Benz vehicles, specifically for the North American market. Unlike traditional cloud-based AI, this partnership emphasizes on-device generative AI for the MBUX (Mercedes-Benz User Experience) system. Key highlights of the partnership include: On-Board Processing: The AI models will run directly on the vehicle’s hardware,

WG bi-weekly meeting April 17 2026

MoM 2026-4-17Download

Gen Edge AI Forum 5 Agenda April 21-22 2026

GenEdgeAI meeting 26-4-2026 MoMDownload

PrismML 1-bit Bonsai LM

Here is a brief executive summary of the article provided by @Mike Stauffer: Core Breakthrough: Caltech-led startup PrismML has developed a mathematical framework for 1-bit large language models, achieving radical compression of high-fidelity AI without sacrificing reasoning performance. Performance Gains: Their flagship Bonsai 8B model reduces memory requirements from 16GB to 1GB, boosts processing speeds by up to 8x,

From Feasibility to Ecosystems:How Generative AI at the Edge Has Evolved

Poster at EdgeAI San Diego 2026 EDGE AI - From Feasibility to Ecosystems How Generative AI at the Edge Has Evolved – RobertoDanilo_TBPDownload

Benchmarking Small Language Models on an Industry-grade, High-end Microcontroller

Presentation at the Research Forum EdgeAI San Diego 2026 EdgeAI_RS_SLM_on_MP2_PAUDownload

Practical Design and Training of Edge Deployable VQA in a Natural Interaction Generative Edge Pipeline

Workshop at EdgeAI 2026 San Diego Workshop VQA Danilo PauDownload

Conversational agents for natural human machine interaction: which type of edge device?

Speech at EdgeAI 2026 San Diego C-AI-EAISD-PAUDownload

From Feasibility to Ecosystems

A Personal Reflection from the Working Group Chairs on the Evolution of Generative AI at the Edge Across the Four Generative Edge AI Forums Two years ago, when we first began discussing generative AI at the edge within the Edge AI Foundation community, the mood was cautiously curious. The dominant

MoM WG meeting 2/27/2026

MoM feb 27 2026 GenEdgeAI WGDownload

Liquid.AI MoE

LFM2-24B-A2B, a 24B Mixture-of-Experts model with 2.3B parameters active per token, built on Liquid AI hybrid, hardware-aware LFM2 architecture.By activating only the most relevant parameters at runtime, LFM2-24B-A2B delivers large-model capability with fast, memory-efficient behavior in a 32GB, 2B-active footprint. https://huggingface.co/LiquidAI/LFM2-24B-A2B-GGUF Check ollama https://ollama.com/search?o=newest

News of the day: Detecting and preventing distillation attacks

Detecting and preventing distillation attacks \ Anthropic Anthropic (2/23) identified industrial-scale distillation (not subject to export control) campaigns by three AI foreigner laboratories.The only way to train edge LM with agentic reasoning ?

Featured paper: Turning a Thermostat into an Agent at the Edge

Abstract Recent advancements in large language models (LLMs) have enabled reasoning-like behavior and sophisticated AI agents, but their deployment usually depends on cloud infrastructure or high-performance hardware. Existing local implementations rarely target energy-efficient edge scenarios. This work presents a conversational AI thermostat assistant running fully on a Raspberry Pi 5,

Featured paper: Artificial Intelligence at the edge: A joint European Roadmap for Edge AI

October 2025 Available at INSIDE and EPoSS Publish Joint European Roadmap for Edge AI - INSIDE Industry Association INSIDE–EPoSS Joint European Roadmap for EdgeDownload

News of the day:

Feature paper: Action Prediction with Edge Generative AI for Mice Pre-clinical Studies

Abstract The well-being of mice is critical in pre-clinical research laboratory studies to achieve high-quality experimental results and to mitigate ethical concerns. In this context, the paper proposes a generative intelligent solution for producing human-readable descriptions of mice’s behavior living in a video sensorized cage. By deploying models on off-the-shelf

Featured paper: Transitioning from TinyML to Edge GenAI: A Review

Abstract Generative AI (GenAI) models are designed to produce realistic and natural data, such as images, audio, or written text. Due to their high computational and memory demands, these models traditionally run on powerful remote compute servers. However, there is growing interest in deploying GenAI models at the edge, on

Generative AI at the edge: challenges and opportunities: the next phase in AI deployment

Authors: Vijay Janapa Reddi 3733702_Download

Generative Edge AI Forums

The first virtual forum : Generative AI and Foundation Models on the Edge . It took place on March 27-28, 2024 Call for presentation @ https://www.edgeaifoundation.org/events/tinyai-virtual-forum-on-generative-ai-and-foundation-models-on-the-edge All speeches available @ https://www.youtube.com/playlist?list=PLeisuBi-nfBPZblK-ik_KLK8tRUlDCBZ3 The second virtual Forum : Beyond LLMs and Chatbots – the Journey to Generative AI at the Edge. It