CodeDocs Vault

Kubiya.ai - Product & Platform

Core Technology / Architecture

Multi-layered orchestration system:

  1. Control Plane -- Coordination hub managing routing, shared config, and policy enforcement. SaaS or self-hosted.
  2. Distributed Task Workers -- Execute workloads in isolated MicroVM containers with streaming logs and full audit trails.
  3. Context Graph ("The Brain") -- Vector-database-backed system ingesting data from Slack, docs, code repos, cloud providers, and identity systems to build a living map of tribal knowledge and infrastructure topology.
  4. Governance & Policy Layer -- OPA-based guardrails with RBAC/ABAC, compliance enforcement, Policy-as-Code.
  5. Unified API Layer -- REST, GraphQL, webhooks, event streaming.
  6. Multi-Model Orchestration -- Supports 100+ LLM providers via LiteLLM. Automatic failover and cost optimization.

Key Architectural Principles

Key Features

Feature Description
AI Teammates / Virtual Team Customizable roles (Manager, Dev, Security, Ops) defined as code
Meta Agent Primary orchestration interface coordinating specialized agents
Domain-Specific Agents Terraform, GitHub, Shell, PagerDuty, Jira, Kubernetes, etc.
Natural Language Interface Plain English via Slack, Teams, or CLI
Context Graph Adaptive recall of logs, past incidents, verified solutions, infra topology
Task Kanban Real-time board tracking Pending/Running/Waiting/Completed/Failed
Human-in-the-Loop Approvals Embedded approve/deny control points
Goal Setting & ROI Tracking Define success metrics and measure engineering productivity
Cognitive Memory Agents learn from execution history; shared across team
Connectors Pre-built: AWS, GitHub, Jira, Slack, Kubernetes, Terraform, CI/CD, custom APIs
MCP Support Model Context Protocol for standardized tool integration
Enterprise Security SOC 2 Type II, GDPR, CCPA; HIPAA-ready for self-hosted; air-gapped support

How It Works (6 Steps)

  1. Sign in at compose.kubiya.ai and launch the Meta Agent
  2. Connect services (AWS, GitHub, Jira, Slack, K8s) via Connectors
  3. Ingest data sources to populate the Context Graph
  4. Query infrastructure through natural language
  5. Execute tasks via Task Queues with isolated MicroVM execution
  6. Scale with specialized agents and teams

Deployment Options

Pricing

AEH (Agentic Engineering Hours) retainer model with yearly commitments:

Plan Cost Key Features
Professional 2,500 AEH/year Full platform, unlimited agents, hosted Context Graph, SSO
Enterprise Custom + Forward Deployed Engineer, BYO LLM, dedicated support, run in your cluster

2-month pilot program available, converting to yearly agreement upon success.

Impact Metrics (Customer-Reported)