Welcome to DODIL
DODIL is building a comprehensive AI cloud — an end‑to‑end platform that gives engineers a single, consistent stack across the full lifecycle of AI applications.
This is the documentation hub. Each product below has its own guides, API reference, and workflows — open the one you need.
K3
Augmented object storage
An augmented S3 platform that keeps standard object-storage semantics, then layers ingestion pipelines and query services on top so the same data can be searched and processed as vectors, structured tables, or source-driven jobs.
- S3-compatible buckets and objects with familiar semantics
- Ingestion pipelines for vector, table, and source-driven data
- Vector (semantic + hybrid) and SQL / HTAP query services
VBase
Vector database control plane
Dodil's control-plane and access layer for Milvus-based vector services. It provisions tenant-scoped clusters and databases, resolves tenant access, forwards commands to Milvus, and manages backups.
- Tenant-scoped provisioning of Milvus cluster & database services
- Access resolution (endpoint, port, database) + unified RunCommand forwarding
- Backup and backup-policy management
Scriptum
Workflow orchestration
A multi-tenant workflow orchestrator for .scriptum DSL programs, with a full authoring lifecycle, durable thread execution, and human-in-the-loop control.
- Authoring lifecycle: draft, compile, test, publish, rollback
- Durable thread execution with step-level telemetry
- Pause / resume for human-in-the-loop (ask) steps and tool invocation
Ignite
Serverless apps & inference
Dodil's serverless execution platform for tenant-scoped apps and managed model inference — from app lifecycle and invocation through to a remote build pipeline and managed inference endpoints.
- App lifecycle: create, draft, compile, test, publish, rollback
- Streaming & async invocation plus a remote container build pipeline
- Managed inference: chat, embeddings, infer, transcribe, rerank