Shelter41
Shelter41 is an infrastructure and control plane for long-running AI coding agents. It gives agents persistent project workspaces, session context, lifecycle controls, observability and governance so teams can coordinate work across runtimes without keeping compute continuously running.
At a Glance
- Software engineering teams and organizations using AI coding agents
- Solo founders and small teams coordinating parallel agent work
- Enterprise engineering organizations requiring private-cloud or on-premises deployment
- Teams using Claude Code, Codex and other coding-agent runtimes
- +1 more
AI Tools by Shelter41
(1)
Shelter41
Cloud Workspaces for Coding Agents
Discussions
No discussions yet
Be the first to start a discussion about Shelter41
Latest News
Shelter41 introduces Swarms, coordinated teams of AI agents for one goal
Shelter41 details parallel agents, local session import, Director View, CLI controls and repository/MCP integrations
Shelter41 publicly presents its beta as a control plane for long-running AI coding agents
Hatchery launches as a private-development product for persistent AI coding engineers with compounding engineering memory
Products & Services
Hosted workspaces for coding agents with persistent files, repository instructions, shared memory, prior conversations and retained session context; compute can sleep while the workspace remains available.
A control plane for assigning work, allocating isolated sandboxes, routing tasks across runtimes, monitoring activity and coordinating agent handoffs. Director View shows running, parked, failed, blocked and waiting agents.
A persistent, isolated workspace for Claude Code with project context, team visibility, usage controls and pause/resume behavior.
A durable project workspace for Codex CLI that retains files, instructions and session context while idle compute can be paused.
Market Position
Shelter41 positions itself as the operational/control-plane layer around coding agents rather than as another chat interface or a single agent runtime. Its stated frame is closer to GitHub Actions, Linear and Datadog for autonomous coding agents: persistent identities and workspaces, lifecycle and compute control, orchestration, traces, usage visibility and governance. It differentiates from disposable session-based coding tools through persistent context and from vendor-locked products through support for multiple runtimes, custom agents and Kubernetes deployment.
Leadership
Founders
Younes Khadraoui
AI systems architect and developer with a PhD in distributed systems/telecommunications; founder of Shelter41 and previously technical co-founder and solutions lead at Namla, a Kubernetes-native cloud and edge orchestration platform. Earlier roles include Lead Machine Learning Engineer at Sanofi Pasteur, Machine Learning Engineer at Mantu, research/software engineering at IMT Atlantique/IRISA and Eurecom, and freelance medical-systems engineering.
Founding Story
The project was started in response to the operational problems that appear when running multiple coding agents across projects and runtimes: tracking what is working, waiting or failed; preserving context between runs; controlling compute; and coordinating Claude Code and Codex without jumping between terminals. Shelter41 was conceived as an operational layer for an agent workforce rather than another chat interface, with the agent treated as a persistent entity attached to a workspace, runtime and repository.
Business Model
Revenue Model
Usage-based hosted Pay As You Go, billed from a shared organizational credit balance for running-agent compute and retained storage; AI-provider subscriptions and model usage are billed separately by the provider. Enterprise deployments are custom agreements covering managed or on-premises deployment, capacity and support.
Pricing Tiers
Starts with $20 in credits; no subscription or seat fees; up to 10 team members, 20 workspaces and 10 simultaneous agents; 1 GiB storage included per organization; credits do not expire.
$20, $60 and $200 correspond to 40, 120 and 400 Standard agent-hours before storage charges.
Managed by Shelter41 or on-premises on the customer's Kubernetes infrastructure, with custom capacity, deployment planning, onboarding, support and invoicing.
Target Markets
- Software engineering teams and organizations using AI coding agents
- Solo founders and small teams coordinating parallel agent work
- Enterprise engineering organizations requiring private-cloud or on-premises deployment
- Teams using Claude Code, Codex and other coding-agent runtimes
- Organizations with Kubernetes infrastructure, network-policy, storage, governance or audit requirements
- Running multiple coding agents across features, bug investigations, tests and documentation in parallel
- Long-running software engineering tasks that span reviews, failures, credentials or human decisions
- Resuming local Claude Code or Codex work in remote cloud workspaces
- Coordinating build, test and review agents around one project goal
- Enterprise self-hosted or private-cloud agent deployments with network, storage and operational controls
- PR reviews, migrations and incident fixes using persistent AI coding engineers