EU AI Act — transparency rules apply Aug 2026, high-risk obligations Dec 2027

Don't ship AI
you can't govern

The open-source governance layer that runs inside your stack — enforce policies locally, keep every prompt in your own infrastructure, and stream a tamper-evident decision log to your dashboard. No proxy. No data egress. No $50K contract.

Get Started Free

Join the waitlist for early access:

Open-source SDKStreaming governanceLangChain integrationZero event loss
terminal — overrule demo
live

Production-ready from day one

140
Tests passing
5
Policy engines
5
Python versions
Sub-ms
Typical latency
0
Data egress
MIT
Licensed
The problem

AI governance is broken

Every option forces a tradeoff that shouldn't exist. Expensive or incomplete. Secure or usable. Fast to deploy or audit-ready.

💸

Enterprise vendors want $50K+ and a 6-month pilot

By the time you're onboarded, you've missed three compliance deadlines.

🔒

Proxy-based tools route your data through their servers

Your security team vetoes it. Your users' data leaves your VPC.

💀

Their platform goes down, your AI goes dark

Governance should protect your product, not become a single point of failure.

📄

Open-source libraries give you detection — but no proof

The auditor doesn't care about your regex. They want a trail you can export.

Overrule eliminates these tradeoffs entirely.

Integration

Three steps. Zero config.

No infrastructure. No sidecars. No proxy. Just import and call.

01

Install

$pip install overrule1.2s
02

Wrap your LLM calls

async with Guard() as guard:
response = await guard.chat(
model="gpt-4o",
messages=messages,
)
03

Every call is governed

pii-detectionPASS
injection-detectionPASS
toxicity-filterBLOCK
event #29471 → audit trail (2ms)
Cloud dashboard

Every AI decision. One screen.

Real-time visibility into what your AI is doing, what policies caught, and what auditors need.

overrule.dev/dashboard

AI Governance Command Center

Security posture is strong today.

Monitor policy enforcement, investigate risk signals, and keep your LLM stack compliant with minimal latency overhead.

Runtime healthy4/5 policies activeRealtime active

Events processed

12,847

+1,204 today30-day window

Violations detected

23

+3 todayAction required

Avg latency

0.8ms

Within budgetPolicy eval only

Fail-open incidents

0

All clearLast 7 days

Starter Plan

$39/mo

View Plan

Events

51%

12,847 / 25,000

API Keys

32%

8 / 25

Data Retention

27%

24 / 90 days

Rate Limit

500 req/min

7-Day Activity

EventsViolations
Mon
Tue
Wed
Thu
Fri
Sat
Sun

Live Enforcement Stream

Live · 47 events
gpt-4oopenai
2s ago
1847 tokens142mspii-detection, injection-detection
!claude-3.5-sonnetanthropic
8s ago
2103 tokens189ms1 violationpii-detection
gpt-4oopenai
14s ago
934 tokens98mstoxicity-detection
gpt-4o-miniopenai
31s ago
512 tokens67msinjection-detection
!claude-3.5-sonnetanthropic
45s ago
3201 tokens234ms1 violationinjection-detection
View all events

Policy Coverage

Endpoints governed

4/5

1 endpoint pending rollout

PII Detection

Active

Injection Guard

Active

Toxicity Filter

Active

Jailbreak Detection

Beta

What you see when you sign up. No sales call required.

Why this architecture wins

Every design decision exists
because alternatives failed here

Not a prototype. Not a wrapper. Not a proxy. Production infrastructure that solves the problems other tools create.

Fail-Open — Governance ≠ Outage

Other platforms go down, your AI goes dark. Overrule fails open. SDK errors never crash your application. Your uptime is sacred, not sacrificed for governance.

In-Process — No Proxy Tax

Policies evaluate locally, in-process — typically sub-millisecond on real-world prompts, with no network round-trip on the hot path. While others add 50-200ms per request routing through their proxy, we never sit between you and your provider.

Audit Trail — Prove, Don't Promise

Every governed call produces a structured decision record: timestamp, model, policies applied, violations with severity, and latency. Your prompts and completions never leave your infrastructure — only governance metadata is streamed.

Multi-Provider — No Lock-In

OpenAI and Anthropic today, more providers coming. Switch between them without touching governance logic. Other tools lock you into their proxy and their provider partnerships.

Your Rules, Your Code — Ship in Minutes

PII, injection, toxicity built-in. Streaming interception, LangChain callback, policy hot-reload. Need custom rules? Subclass BasePolicy. No tickets, no vendor timelines.

Battle-Tested Resilience

Circuit breakers, exponential backoff, bounded buffers, graceful shutdown. Production infrastructure with fail-open resilience from day one.

Why Overrule

Enterprise governance without
the enterprise baggage

Existing solutions force you to choose: expensive platform with a 6-month rollout, or fragile DIY scripts with no audit trail. Overrule is the third option.

5 min
To production
Not 5 months. Install, wrap, enforce.
100%
Open-source SDK
Audit every line. No black boxes.
$0
To start
No sales call. No credit card.
0 ms
Added failure risk
Fail-open. Your uptime is sacred.
Dimension
Traditional vendors
Overrule
Getting started
6-month sales cycle, SOW, pilot program
pip install overrule — production in 5 minutes
Pricing
$50K+ annual contracts, per-seat licensing
Free tier forever. Scale at $39–$149/mo
Architecture
Closed SaaS proxy — your data routes through their servers
Open-source SDK runs in your infra. You own the data path
Failure mode
Platform outage = your AI goes down
Fail-open by default. Governance never becomes an outage
Compliance proof
Static policy docs and quarterly audits
Continuous audit trail. Every decision logged, exportable, Article 13 ready
Customization
Vendor-controlled policy catalog, change requests take weeks
Subclass BasePolicy — ship custom rules in minutes, not months

Overrule is what happens when you build governance for developers, not for procurement teams.

EU AI Act — Transparency rules apply Aug 2026 • High-risk obligations by Dec 2027

Compliance infrastructure
doesn't build itself
the week before a deadline.

The EU AI Act's high-risk obligations (Articles 13-15) were deferred to December 2027 under the Digital Omnibus — but transparency rules under Article 50 apply from August 2026, and the logging infrastructure you'll need doesn't appear overnight.

Overrule gives you runtime enforcement and a real decision trail now, so compliance is a switch you flip — not a fire drill.

What Overrule provides
  • Structured decision log for every governed AI call
  • Policy enforcement provably active at all times
  • Prompts and completions never leave your infrastructure
  • Exportable metadata + violation evidence for regulators
  • Real-time violation detection and alerting
  • Governance records mapped to EU AI Act articles
Pricing

Start free. Scale when it matters.

No credit card required. The SDK is open-source and free forever. Upgrade when you need more capacity.

Stop prompt-injection and unsafe outputs in real time
Protect customer data with policy-driven controls
Prove compliance with immutable audit trails
Scale safely with usage-based economics

Free

$0forever

For developers getting started with AI governance.

  • 10K requests/month
  • All policies incl. custom
  • 5 API keys · 1 webhook
  • 7-day log retention
  • 200 req/min rate limit
  • Community support
Start Free
Most Popular

Growth

$499/mo

For production AI apps that need full coverage.

  • 1M requests/month
  • All policies incl. custom
  • 25 API keys · 5 webhooks
  • 2,000 req/min rate limit
  • 30-day audit trail
  • Email support

Scale

$2,999/mo

For high-volume, compliance-sensitive teams.

  • 10M requests/month
  • Everything in Growth
  • 100 API keys · 20 webhooks
  • 10,000 req/min rate limit
  • 90-day log retention
  • SSO/SAML · 99.9% SLA
  • Priority support + shared Slack
Contact Sales

Enterprise

$10k+/mo

For regulated and mission-critical deployments.

  • Unlimited volume
  • Dedicated cluster / private deployment
  • 365-day retention + data export
  • 50,000 req/min rate limit
  • Legal-ready audit exports
  • Dedicated success & security support
Contact Sales

The open-source SDK is free forever. Fail-open architecture and sub-millisecond policy evaluation included in all plans.

FAQ

Answers to the objections every AI team asks

Stop choosing between
fast and compliant

Open-source SDK. Cloud audit trail. $0 to start.Enterprise governance without enterprise procurement.

MIT LicensedNo vendor lock-inSelf-host or cloud
Try DemoGet Started