AI Engines
Hosting Compatibility
Owlpen can be deployed across major managed cloud platforms, self-managed infrastructure, and fully air-gapped on-premises environments.
Compatibility ratings reflect tested deployment configurations. Actual performance may vary based on instance sizing and network configuration.
Managed Cloud
| Provider / Service | Description | Owlpen |
|---|---|---|
| AWS BedrockNoteNative model integrations via Bedrock APIs; Longbow requires custom agent configuration.aws.amazon.com | Amazon's fully managed AI service. Run Longbow and Ledger workloads using Bedrock's agent and model APIs. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Google Vertex AINoteVertex AI Pipelines supported for Ledger; Longbow partial via custom containers.cloud.google.com | Google's managed ML platform. Deploy Ledger workloads with Vertex AI endpoints and pipelines. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Azure AI StudioNoteLongbow and Ledger require additional container configuration.ai.azure.com | Microsoft's managed AI development platform. Both platforms deployable via Azure AI managed endpoints. | StandardStandardStandard support. Most tasks handled reliably; some advanced, context-heavy, or multi-step operations may produce degraded results. |
| AWS SageMakerNoteLongbow's agentic workflows have limited SageMaker support.aws.amazon.com | Amazon's ML operations platform. Host Ledger as SageMaker endpoints with auto-scaling. | StandardStandardStandard support. Most tasks handled reliably; some advanced, context-heavy, or multi-step operations may produce degraded results. |
| Hugging Face Inference EndpointsNoteUsed to serve open-source models to Ledger, not to host the platforms directly.huggingface.co | Managed model hosting. Suitable for self-hosted open models used within platform integrations. | StandardStandardStandard support. Most tasks handled reliably; some advanced, context-heavy, or multi-step operations may produce degraded results. |
Self-Managed Cloud
| Provider / Service | Description | Owlpen |
|---|---|---|
| AWS (EC2 / ECS / EKS)aws.amazon.com | Amazon's compute and container services. Full platform support via Docker containers and Kubernetes. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Google Cloud (GKE / Cloud Run)cloud.google.com | Google's container and serverless platforms. Both platforms deploy cleanly via GKE or Cloud Run. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Azure (AKS / Container Apps)azure.microsoft.com | Microsoft's Kubernetes and serverless container services. Full deployment support via Docker and Helm. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| DigitalOcean (Droplets / DOKS)NoteResource limits on smaller droplets may affect Longbow performance at scale.digitalocean.com | Developer-focused cloud with managed Kubernetes. Good for smaller-scale or development deployments. | StandardStandardStandard support. Most tasks handled reliably; some advanced, context-heavy, or multi-step operations may produce degraded results. |
| Hetzner (Dedicated / VPS)hetzner.com | European cloud and dedicated server provider. Cost-effective self-managed hosting via Docker or K3s. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Fly.ioNoteResource constraints limit Ledger's data pipeline capabilities.fly.io | Edge-native container platform. Suitable for lightweight Longbow deployments close to users. | LimitedLimitedLimited support. Basic functions only. Complex reasoning, extended context, and multi-step workflows will degrade significantly. |
| RailwayNoteLedger's storage requirements exceed Railway's current tier limits.railway.app | Simple container deployment platform. Good for staging or low-traffic Longbow instances. | — |
On-Premises & Air-Gapped
| Provider / Service | Description | Owlpen |
|---|---|---|
| Linux (Ubuntu / Debian / RHEL)ubuntu.com | Primary recommended OS for all self-hosted deployments. Full support via Docker, systemd, or bare-metal install. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Windows ServerNoteLongbow's subprocess management has minor overhead on Windows; WSL2 recommended.microsoft.com | Windows Server 2019+ supported via Docker Desktop or WSL2. Some performance overhead compared to Linux. | StandardStandardStandard support. Most tasks handled reliably; some advanced, context-heavy, or multi-step operations may produce degraded results. |
| Kubernetes (self-managed)kubernetes.io | Self-managed Kubernetes clusters (K3s, RKE, kubeadm). Recommended for production on-premises deployments. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| Docker (standalone)docker.com | Docker Compose deployments. Recommended for single-server or development environments. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
| VMware vSphere / ESXiNoteCPU scheduling overhead on ESXi may affect Longbow latency.vmware.com | Enterprise virtualisation. Run platform containers inside VMs with standard Docker or Kubernetes. | StandardStandardStandard support. Most tasks handled reliably; some advanced, context-heavy, or multi-step operations may produce degraded results. |
| Air-gapped / OfflineNoteLedger fully supported offline. Longbow requires additional configuration for local model routing.coaleypeak.co.uk | Fully disconnected deployments with no internet access. BYOK not available; requires pre-loaded model weights. | FullFullFull support. All platform features available with reliable output quality, including complex reasoning, multi-step workflows, and large context tasks. |
Hosting compatibility is subject to change as cloud provider APIs and infrastructure offerings evolve. Configuration requirements for each platform are documented in the deployment guides. All service names, trademarks, and logos are the property of their respective owners. Coaley Peak® Ltd is not affiliated with or endorsed by any third-party hosting provider.
← Back to About