The Governance Proxy for AI Agents
SDK wrappers are door locks.
Your agent can still find another door.
Govyn is a wall. No API key = no way through.
Explore Govyn
Integrations, policy templates, and insights for governing AI agents in production.
Add budget limits, policy enforcement, and full replay to LangChain agents using OpenAI. Five-minute setup, zero code changes.
INTEGRATIONAdd budget controls and policy enforcement to CrewAI multi-agent crews using OpenAI. Govern every agent in your crew independently.
INTEGRATIONGovern OpenClaw agents using Claude. Add budget enforcement, model policies, and conversation replay to your OpenClaw workflows.
FROM OUR BLOGHow smart model routing through a proxy cut our OpenAI and Anthropic bill from $2,140/mo to $578/mo. Zero code changes. Just YAML.
Door Lock
Wall
How it works
Install & configure
One command to start. Point it at your providers.
$ npx govyn init
# config.yaml
providers:
openai:
api_key_env: OPENAI_API_KEY
agents:
research-agent:
budget_daily: 50.00Define policies
YAML rules. Git-versioned. Enforced at the proxy.
policies:
- name: "no-database-deletes"
scope: global
rule: block
match:
content_pattern:
"DELETE FROM|DROP TABLE"
message:
"Blocked by policy"Your agents are governed
Every call logged. Budgets enforced. Policies can't be bypassed.
Before
After
What no one else has
Proxy enforcement
Not an SDK. Your agent never has the real API key. Architecturally unbypassable.
Policy-as-code
YAML policies versioned in git. Not just budget numbers in a dashboard.
Cost control & budgets
Per-agent budgets, loop detection, auto-kill. Cost tracking across all providers.
Smart model routing
Route to cheaper models automatically. Save 60-80% on tokens with zero code changes.
Session replay
Record and replay full agent sessions. Debug what happened at 3am.
5-minute setup
npx govyn init. YAML config. No Kubernetes. No enterprise onboarding.
Cut your AI API bill by 60-80%
without changing agent code
- name: smart-model-routing
type: model_route
rules:
# Short prompts → Haiku
- when:
input_tokens_estimate: "<500"
route_to: "claude-haiku-4-5"
# Medium → Sonnet
- when:
input_tokens_estimate: "<4000"
route_to: "claude-sonnet-4-5"
# Complex → stays as requested
- default: passthroughCost per million input tokens
Your agent requests Opus → Proxy sends Haiku → You save ~$14/M tokens
70% of agent calls are simple tasks that don't need a premium model.
How Govyn compares
| Feature | Govyn | SDK Wrappers AgentBudget, TealTiger | Enterprise Agentgateway |
|---|---|---|---|
| Architecture | HTTP proxy | In-process SDK wrapper | K8s-native proxy |
| API key location | Proxy only | Agent environment | K8s secrets |
| Bypass possible? | No | Yes | No |
| Policy engine | YAML, git-versioned | Built-in rules only | Enterprise config |
| Cost tracking | Built-in, per-agent | Built-in, basic | Enterprise tools |
| Setup time | 5 minutes (npx) | 3 lines of code | Enterprise onboarding |
Pricing
Free / Open Source
Self-hosted, forever
- Unlimited agents
- 24-hour log retention
- Basic policies
- Community support
Starter
Cloud-hosted proxy
- 10 agents
- 7-day log retention
- Smart model routing
- Email support
Team
For production teams
- Unlimited agents
- 30-day log retention
- Session replay
- Approval queue & RBAC
- Priority support
Need more? Enterprise plans coming soon.
Get early access
Join the waitlist for Govyn Cloud. We'll notify you when it's ready.
No spam. We'll only email you when there's something to show.
You're on the list.
We'll reach out when Govyn Cloud is ready.
Something went wrong. Please try again.