Coder Agent Relay: run Cursor agents inside your network

On September 3, 2026, Coder announced Agent Relay — a self-hosted execution environment for cloud coding agents, with SpaceXAI as its launch partner. Cursor's cloud agents can now run inside Coder workspaces on infrastructure the customer already operates: developers keep the Cursor experience they know, but tool calls execute on the customer's network, so source code, secrets, and internal services stay on machines they control.

A market that AI coding could never reach

Demand for AI coding agents inside large enterprises has exploded, but adoption in regulated industries hasn't kept pace — not for the reason most people assume. Developers in these organizations want the same tools everyone else has. The blocker is the deployment model. Banks, defense agencies, government institutions, and global enterprises operate under requirements vendor-hosted tools can't satisfy: source code needs controlled access, execution environments need security and governance that prevent data exfiltration, and every action needs to be auditable.

Gartner projects that 80% of enterprise software engineers will need to upskill for generative AI by 2027. In the most regulated sectors, security and compliance review still decide which tools ever reach a developer — which is exactly the market Coder opens. Coder is the self-hosted infrastructure the world's most demanding organizations already trust to run development inside their own walls. Pairing that foundation with SpaceXAI's agentic experience turns a market that has been closed to vendor-hosted AI tools into one the two companies can finally enter together.

AI needs an operating layer

The modern enterprise stack runs in layers: application, data, compute. AI fits none of them. It touches all three and belongs to none — which is why organizations everywhere hit the same wall the moment AI moves from pilots into production. Coder calls the answer the AI Operating Layer: the layer where AI runs inside the organization, on its own infrastructure, under its own policies, visible to its own teams. Built right, it enforces five things — data stays inside the boundary, every task sees only what it was granted, only approved models run, every action leaves a record, and spend is capped before it accumulates.

Coder CEO Rob Whiteley put it bluntly: "Enterprises never rejected AI agents. They rejected the deployment model. The model was never the bottleneck. What has kept this work out of production is everything that comes after the pilot: where the code runs, who has access, what the agent can reach, and what record exists afterward. Those are infrastructure questions, and Coder was built to answer them."

What Coder Agent Relay delivers

Each Coder workspace can start a Cursor worker that opens an outbound connection to Cursor. Platform teams can provision and scale those workspaces the same way they already manage developer environments, so regulated organizations can deploy cloud agents at scale next to private code and custom hardware without giving up Cursor's product surface or the security and standardization of Coder workspaces.

Four principles are worth unpacking. First, your perimeter, your rules: workspaces run on the customer's cloud, VPC, or on-premises environment; code does not leave; network egress policy is set once by the platform team and enforced uniformly across every workspace. Second, boundaries by architecture, not by trust: agent environments are sandboxed, ephemeral, and scoped to a single task — a prompt injection pushing an agent toward unauthorized resources is blocked at the environment layer, not left to the model to refuse. Third, policy as infrastructure: approved models, permitted data sources, and available resources are defined once at the environment level and inherited by every workspace. Fourth, a complete record: every run produces a log of what the agent accessed, executed, changed, and was blocked from doing — compliance reporting for any time window does not have to be reconstructed by hand.

What it means for regulated industries

Toni Adams, Senior Director of Partnerships at SpaceXAI, captured the tension on both sides of the market: "Engineering teams want to build with Cursor. Enterprise security policies require that sensitive information not leave its own infrastructure. Coder provides those teams a path to run Cursor inside environments they fully control, giving developers the experience they want while maintaining the security controls they need."

For governance decision-makers, the real signal is not one product but a shift in framing: the governance question for coding agents has moved from "which model is approved" to "where do the tool calls execute, who has access, and what record exists afterward." Agent Relay is the infrastructure-layer answer to the lesson vendor-hosted coding tools taught us — execution location itself is a governance decision.

Limits and what comes next

Agent Relay solves execution location, not all of agent governance: prompt injection is blocked at the environment layer, but an agent's decision quality, goal alignment, and cross-session persistence remain separate problems. SpaceXAI is the first provider with Agent Relay, and Coder says it is extending self-hosted execution to more agent providers over time — the signal for enterprises is that "self-hosted execution" is moving from exception to standard option.

References: AiThority / GlobeNewswire, "Coder and SpaceXAI Collaborate to Bring Agentic Coding Inside the World's Most Regulated Enterprises" (2026-09-03) — https://aithority.com/machine-learning/coder-and-spacexai-collaborate-to-bring-agentic-coding-inside-the-worlds-most-regulated-enterprises/ ; Coder — https://coder.com/