Langdock GmbH
Weighted rubric averages. No axis score is stored in YAML.
Configurability
3.82
out of 5.00
Enterprise readiness
3.36
out of 5.00
Every score, reached definition, rationale, and source.
Admins can add catalog or manual custom model endpoints, control workspace visibility and agent availability, and configure routed fallback deployments.
Organizations can connect their own model endpoints and define routing or availability policies.
Agents support instructions, knowledge, integrations, workflows, subagents, granular sharing, drafts and publishing; admins can review, approve, flag, or disable them.
Organization-managed agents support permissions, sharing, approved data, and controlled tools.
A parent agent can delegate tasks to specialized subagents that run with their own instructions, knowledge, actions, and model.
Agents can delegate to or invoke other specialized agents.
Langdock connects remote MCP servers over HTTP or SSE with API-key or OAuth authentication, discovers tools, and lets admins restrict integrations and actions to users or groups.
Supports authenticated remote tool servers with organization administration.
Workflows support external or form triggers, multi-step actions, conditions, loops, human approval, API calls, and workspace controls for risky node types.
Event-driven workflows support branching, approvals, and external systems.
Authenticated APIs cover model completion, agents, knowledge bases, integrations, skills, usage export, user management, and audit logs.
Enterprise APIs cover agents, tools, knowledge, administration, or automation with service authentication.
Company Knowledge and synced folders provide permission-aware multi-source retrieval, citations, daily refresh, source access checks, and admin-controlled availability.
Permission-aware enterprise sources support sync, citations, and admin controls.
Agent Evals runs structured test sets before publishing and supports configurable AI-judge, tool-call, and keyword checks with recorded runs.
Automated evaluations compare prompts, agents, models, or releases with configurable graders.
The prompt library supports reusable prompts, variables, folders, and sharing with teams or the workspace.
System instructions, variables, and reusable organization templates.
Developers can build and manage custom integrations with authentication, actions, triggers, and sandboxed JavaScript, then administer their distribution across a workspace.
Developers can package, distribute, and administer extensions across an organization.
Langdock documents multi-tenant and single-tenant SaaS, deployment in a customer's cloud, and on-premises Kubernetes deployment via Helm charts.
Customer cloud, VPC, or self-hosted deployment is supported.
Langdock supports SAML 2.0 with compatible identity providers and enforces SSO for verified workspace domains.
Standards-based SAML or OIDC SSO for enterprise workspaces.
SCIM 2.0 automates provisioning, updates, deactivation, deletion, and synchronization of Microsoft Entra users and groups.
SCIM groups and role or team mapping with automated lifecycle management.
Admins can customize permissions for Member, Editor, and Admin roles across agents, workflows, knowledge, skills, integrations, projects, prompts, groups, and individual actions.
Custom roles or granular permissions cover data, agents, tools, and administration.
The filterable Audit Logs API records actors, actions, entities, IP and user-agent data, before/after changes, and broad workspace configuration events for 90 days.
Searchable admin audit logs cover key user and configuration events.
Langdock documents ISO 27001 certification, SOC 2 Type II audit, GDPR-aligned processing, a DPA, and independent penetration testing.
SOC 2 Type II or ISO 27001 plus GDPR-aligned enterprise terms.
The SaaS application is hosted in the EU; model regions are configurable, and dedicated customer-cloud or on-premises deployments allow customer-controlled location.
Customer-selectable storage region for primary product data.
Admins control models, product access, data retention, tools, and sharing; Governance adds agent review plus automated compliance checks on every published version.
Admin policies control models, sharing, data sources, or tools.
The admin workspace manages users, groups, domains, roles, security, models, capabilities, product access, retention, integrations, workflows, and delegated Editor permissions.
Delegated administration, groups, domains, policies, and bulk controls are supported.
Admins can export up to 12 months of user, agent, API-key, project, model, cost, and performance data through the UI or scoped CSV/JSON API.
Export or API supports custom reporting, cost allocation, and trend analysis.
Enterprise customers receive custom support and a shared channel with direct Langdock access, coordinated through an account manager.
Enterprise support plan with defined response targets or dedicated onboarding.
Langdock documents no model training, configurable chat retention, encryption at rest and in transit, IP allowlists, security administration, and dedicated deployment options.
Enterprise controls include retention, training-data commitments, and security administration.
For each axis: sum(score × rubric weight) ÷ sum(weights). Any result without verified evidence is constrained to score 0 by schema.