Nexilume AI
Nexilume AI builds an edge-first execution fabric for AI agents. Its Nexus platform lets developers host agents in cloud, edge, or their own runtime, route calls across environments, and let users authorize computers and phones for agent work.
At a Glance
- AI-agent developers and application builders
- Organizations operating shared or commercial agent services
- Developer and operations teams
- Enterprises needing private-network, edge, or device-aware agent execution
- +1 more
AI Tools by Nexilume AI
(1)
Nexus Cloud
Python Agent Execution Platform
Discussions
No discussions yet
Be the first to start a discussion about Nexilume AI
Latest News
Nexus Cloud source update adds Relay MCP streaming, agent-training Data Assets, Provider synchronization, and improved stream/error handling.
Nexus Cloud v2026.10.17-beta.2 listed as a GitHub pre-release.
Nexus Cloud v2026.10.05-beta.1 announced as the first tagged beta of the public Cloud + Web source tree.
Nexilume technical report 'Nexus: An Execution Fabric for AI Agents Across Cloud, Edge, and Devices' published on arXiv.
Products & Services
Cloud execution platform for hosting, discovering, invoking, and observing AI agents, with model routing, Private Display, runs, files/Data Assets, device integrations, and marketplace workflows in the commercial edition.
Apache-2.0-modified source-available single-owner self-hosted edition containing Cloud Server, Web Console, and Relay source.
Organization-oriented hosted/commercial edition with tenant and member access, roles, policies, billing, and marketplace capabilities.
OpenWrt-based private, agent-aware edge router for discovering, isolating, policy-routing, and auditing private agents, models, and tools.
Market Position
Nexilume positions Nexus as an execution fabric rather than only a chat application: it separates where an agent is deployed from where user-authorized work occurs, and combines cloud, edge routing, device authorization, persistent execution, and model routing. Its own documentation compares the user model with OpenClaw and Codex, while its technical report evaluates Nexus against Dify for agent workflows and Dapr for service invocation. Its differentiation is cross-environment execution with explicit authorization and run continuity; Dapr has lower successful-call latency in the reported service benchmark, so Nexus is not positioned as universally faster.
Leadership
Founders
Cary Chang
Founder of Nexilume AI and maker of Nexus. The company's technical report credits Chang with conceptualization, system design, and manuscript preparation; no prior employer or company is identified in the sources reviewed.
Executive Team
Cary Chang
Founder
Founder and maker of Nexus; credited with conceptualization, system design, and manuscript preparation for the Nexus technical report.
Founding Story
Cary Chang describes Nexus as starting from the problem that the computer running an agent is often not the computer where the user needs the work done. The initial vision was to separate agent deployment from user-authorized execution environments and connect them through a shared execution context.
Business Model
Revenue Model
The public NexusCloud listing labels the product Freemium. Nexilume's official materials distinguish a single-owner self-hosted Community edition from a commercial Enterprise edition; Enterprise includes organization access, hosted services, marketplace workflows, billing, and TokenBank/commercial accounting. Agent Marketplace listings can use fixed or agent-reported pricing.
Pricing Tiers
Single-owner self-hosted edition; billing, organization administration, and marketplace workflows are excluded.
Organization and commercial operations, hosted services, roles/policies, billing, TokenBank, and marketplace capabilities.
Target Markets
- AI-agent developers and application builders
- Organizations operating shared or commercial agent services
- Developer and operations teams
- Enterprises needing private-network, edge, or device-aware agent execution
- Self-hosters and open-source/community infrastructure operators
- File assistants that analyze spreadsheets or create documents on a user's authorized computer
- Browser automation on a user's attached computer
- Computer-to-phone workflows
- Developer and operations assistants for approved workspaces and logs
- Edge-hosted agent services behind private networks or NAT
- Shared multi-user agent products