EdgeRunner AI

Domain-specific AI for the warfighter

Newsroom

Can Editing 1 Neuron Fix Repetition Loops in LLMs?

Read our new paper here: https://arxiv.org/abs/2606.13705 Download the models here: E2B, E4B, 26A4B, 31B The inner workings of large language models (LLMs) are notoriously difficult to interpret because of their connectedness. Along one dimension, each token in a sequence can influence the others through the self-attention mechanism, and similarly all elements of each embedding are connected across the multi-layer perceptron in each layer. But what if a specific behavior could be localized to

Key Benefits.

Operate Anywhere

EdgeRunner runs directly on laptops, mobile devices, vehicles, and edge hardware - with or without connectivity - enabling AI in disconnected, remote, and contested environments.

Operate Anywhere

EdgeRunner runs directly on laptops, mobile devices, vehicles, and edge hardware - with or without connectivity - enabling AI in disconnected, remote, and contested environments.

Data Security

Sensitive or proprietary data remains local. EdgeRunner eliminates the need to send information to external cloud services, reducing risk of exposure, interception, or compliance breaches.

Data Security

Sensitive or proprietary data remains local. EdgeRunner eliminates the need to send information to external cloud services, reducing risk of exposure, interception, or compliance breaches.

Mission Adaptability

Models are adapted to specific operational domains, workflows, and data sources, delivering relevant outputs aligned to real-world tasks.

Mission Adaptability

Models are adapted to specific operational domains, workflows, and data sources, delivering relevant outputs aligned to real-world tasks.

Near Zero Latency

Local execution removes network delays, enabling rapid responses and real-time decision support when timing matters most.

Near Zero Latency

Local execution removes network delays, enabling rapid responses and real-time decision support when timing matters most.

Hardware Agnostic

EdgeRunner integrates with current hardware and secure environments, minimizing deployment friction and reducing reliance on new infrastructure.

Hardware Agnostic

EdgeRunner integrates with current hardware and secure environments, minimizing deployment friction and reducing reliance on new infrastructure.

Bring Your Own Models

Deploy and run third-party or government-developed AI models within the EdgeRunner platform, enabling a unified, secure runtime environment without dependence on external cloud infrastructure.

Bring Your Own Models

Deploy and run third-party or government-developed AI models within the EdgeRunner platform, enabling a unified, secure runtime environment without dependence on external cloud infrastructure.

Introducing

EdgeRunner

Your personalized, on-device Generative AI assistant for increased productivity and operational effectiveness

Our Vision.

Domain-Specific: We transform private data to create personalized, domain-specific AI assistants, accelerating accurate decision making.

Agentic: We build agents that can take actions and complete tasks on behalf of the user, streamlining workflows and increasing productivity.

On-Device: We develop on-device, air-gapped Generative AI technology that runs without requiring access to the cloud or internet, enhancing data privacy and security.

Industries.

Run your AI locally, securely, and sustainably.

Deploying AI at the edge for the defense and intelligence community can significantly enhance real-time analysis, decision-making, operational efficiency, and security, empowering defense personnel to operate more effectively in both combat and intelligence-gathering environments.