Ship policy changes without fear. Test in production, rollback instantly.
The policy enforcement infrastructure you'd build yourself — deterministic evaluation, A/B testing, HITL workflows — without the 18-month timeline. Self-hosted only.
If you're enforcing policies (content moderation, AI guardrails, fraud detection), you're doing one of:
- Hardcoding logic — policy changes require deploys, no safe testing, debugging is guesswork
- Using SaaS — data leaves your infrastructure, vendor lock-in, limited customization
- Building from scratch — 12-18 months minimum, still missing versioning/testing/audit infrastructure
The core challenge: How do you validate policy changes without breaking production?
Same event + same policy + same state = same verdict. Always. Essential for compliance audits, debugging, and safe testing.
Test new policies in production without affecting users. Split traffic between versions. Compare metrics. Roll back instantly.
Track entity state across events — trust scores, violation counts, rate limits. ACID transactions, replay-safe.
Every decision traced with signals computed, rules matched, and state changes applied. Explain any historical decision.
| Capability | Swiftward | OPA/Cedar | Classification SaaS | Build-It-Yourself |
|---|---|---|---|---|
| Deterministic Replay | Guaranteed | Yes | No | Hard to achieve |
| A/B Testing | Built-in | None | Limited | 6+ months |
| Shadow Mode | Built-in | None | Some | Build yourself |
| Entity State | Built-in | External | Limited | Build yourself |
| HITL Workflows | Built-in | None | Varies | Build yourself |
| Deployment | Self-hosted | Self-hosted | SaaS | Your infra |
| Time to Value | Days | Weeks | Days | 12-18 months |
Live demos (no build required):
| Demo | Description | Time |
|---|---|---|
| Minimal Demo | Simplest setup - perfect for first-time users | 2 min |
| Policy Enforcement Demo | A/B testing, shadow mode, HITL workflows, LLM classification | 5 min (+ ~20 min first run to download ML models) |
Both demos use pre-built images - just run docker compose up.
- Stage: Seeking first design partners
- Looking for: 1-2 design partners to validate approach
- Maturity: Core works, production hardening in progress
What works: Event processing, policy evaluation, state management, A/B testing, HITL queues, audit trails.
What we're validating: Real-world policy complexity, DSL ergonomics, integration patterns.
We're looking for 1-2 design partners to validate Swiftward together. Direct founder access, real use cases, honest feedback.
Email: hello@swiftward.dev — 30-minute scoping call, we'll figure out if there's a fit.
This is the public examples and documentation repository.
- Core source code → Available to customers and partners
- This repo → Live demo, examples, and documentation to evaluate Swiftward
- License → Source-available, see LICENSE
- What is Swiftward? — Core concepts
- How It Works — Architecture overview
- Policy DSL Preview — Writing policies
- Comparisons — vs alternatives
Ready for production? Core works, good for pilots. Production hardening in progress.
Different from OPA? OPA is stateless authorization. Swiftward is stateful policy enforcement with A/B testing and HITL.
Performance? Thousands of events/sec single node, scales horizontally.
Email: hello@swiftward.dev
Founder: Konstantin Trunin