Immutable versions
Published workflow versions remain fixed while runs retain their exact version identity.
PRIVATE · AI ORCHESTRATION
AI Fabric
An independently implemented enterprise automation and AI orchestration baseline centered on a durable workflow core.
Private productExecutable baselineWorkflow enginePython + SQLite
The problem
Why this project exists
Automation becomes hard to reason about when workflow definitions, execution state and retries are mixed together. AI Fabric establishes explicit version and run boundaries.
Verified capabilities
What is actually implemented
These capability descriptions are intentionally limited to the evidence-backed implementation surface.
Idempotent starts
Tenant-scoped uniqueness protects duplicate workflow triggers.
Run history
Runs and step attempts are persisted so execution history is explicit.
Workflow nodes
Deterministic transform, filter, loop and registered action nodes are supported.
Retries and cancellation
Action retries and between-node cancellation are explicit execution semantics.
Replay
Replay creates a new run from stored input and the exact workflow version.
Architecture
How the system is structured
The diagram below represents the current verified architecture boundary; planned or environment-dependent components are described in the engineering notes rather than silently promoted.
Published Workflow
↓
Version Boundary
↓
Durable Run
↓
Step Attempts
↓
Deterministic Nodes
↓
Action Boundary
Evidence and authority
A run stores its exact version ID, tenant, input, output, lifecycle state and idempotency key.
PostgreSQL production durability remains a deployment target rather than a silently claimed fact.
Engineering deep dive
Decisions, trade-offs and boundaries
Technology names matter less than the engineering contracts they enforce.
A run stores its exact version ID, tenant, input, output, lifecycle state and idempotency key.
The baseline is explicitly at-least-once; exactly-once side effects are not claimed.
Cancellation is checked between nodes rather than treated as an opaque process kill.
PostgreSQL production durability remains a deployment target rather than a silently claimed fact.
Project status
Private · evidence-backed status
Executable private baseline. Core workflow semantics are verified; remaining production durability/platform epics are explicitly not represented as complete.
ClassificationPrivate product / engineering case study
EvidenceRepository implementation, tests and project records cross-checked
ClaimsCurrent, planned and limitations kept separate
DeploymentNot changed by portfolio integration
PRIVATE PRODUCT
Architecture can be discussed without exposing the implementation.
Private implementation — architecture and engineering decisions can be discussed.