Now in beta · Fully managed · Provider-agnostic

The managed control plane
for your AI agents.

Policy enforcement, audit trails, and human-in-the-loop approvals — hosted and managed for you. Connect your agents in minutes. Ship with confidence.

Get Early Access Book a Demo

Slack

the-butler

Jordan Today at 9:14 AM

hey @Butler — Jenny just started today. can you give her read-only access to the billing agent?

Butler Today at 9:14 AM APP

🔑 Provisioning New Hire — Jenny

📧 jenny@acmecorp.com

🤖 billing-agent → read-only viewer

🚫 No write, send, or modify access

Butler Today at 9:15 AM APP

✅ Done. jenny@acmecorp.com now has read-only access to billing-agent. Invite sent. Audit log updated.

Jordan Today at 2:01 PM

can you send an invoice to Acme Corp

billing-agent Today at 2:01 PM APP

Sure! A few quick things — what's the amount, what product or service is this for, and which email should I send it to?

Jordan Today at 2:02 PM

$4,200 for Q2 consulting retainer, send to billing@acmecorp.com

billing-agent Today at 2:03 PM APP

✅ Invoice Sent — INV-0142

📬 billing@acmecorp.com · Q2 Consulting Retainer

💰 $4,200.00 · Due May 17 · Payment link included

Jordan Today at 2:04 PM

let me know if they haven't paid by the 20th

billing-agent Today at 2:04 PM APP

Got it — I'll ping you on May 20 if INV-0142 is still unpaid. ✅

billing-agent May 20 · 9:00 AM APP

⚠️ Invoice Still Unpaid — INV-0142

Acme Corp hasn't paid $4,200. Was due May 17 — now 3 days overdue.

Jordan May 20 · 9:02 AM

send them a reminder

billing-agent May 20 · 9:02 AM APP

✅ Reminder sent to billing@acmecorp.com with the payment link. I'll follow up again in 48h if still unpaid.

Jenny Today at 11:08 AM

hey — TechStartup was double-charged on their last invoice. can you issue them a refund for $1,800?

billing-agent Today at 11:08 AM APP

Refund logged. Refunds over $500 require manager approval — I'm sending this to Jordan now.

💰 $1,800 · TechStartup Inc · reason: double charge

billing-agent Today at 11:08 AM APP

Sorry Jenny — your account doesn't have permission to issue refunds. Want me to escalate this to Jordan for approval?

Jenny Today at 11:09 AM

yes please

billing-agent Today at 11:09 AM APP

Done — I've sent the request to Jordan via Butler. I'll let you know once he responds. ✅

Butler Today at 11:09 AM APP

🔑 Refund Approval Required

Jenny has requested a $1,800 refund to TechStartup Inc.

💬 "double charged on last invoice" · requested by jenny@acmecorp.com

Jordan Today at 11:11 AM

approved ✓

👍 1

Butler Today at 11:11 AM APP

✅ Refund approved. billing-agent is processing $1,800 to TechStartup Inc. Jenny has been notified. Audit log updated.

No credits

pay for usage only,
no seat fees or minimums

BYO plan

bring your own
AI subscription

Cancel anytime

no lock-in, no contracts,
export your data anytime

see it in action

Watch Sentrely govern an agent — live

Every action your agents take flows through one chokepoint — logged, policy-checked, and held for your approval when it matters. This is the real thing, running below.

// audit log live

// approval required

billing-agent requests:

stripe:Refund · $5,000

reason: customer dispute #8821 — over your $1k auto-limit

// guardrail · charter

support-bot asked db-agent to delete a customer record.

⛔ blocked — charter:
"read-only: answer questions, no writes"

↑ approve or deny — the agent unblocks and the decision is logged, immutably.

AI agent stories, every 2 weeks

Real lessons on running AI agents in production. No spam.

supported ai providers

Bring your own model. Provider-agnostic by design.

Connect via API key or use OAuth with your favorite AI clients (Claude Code, Codex). Mix providers, set fallbacks, and keep agents running even when one goes down.

API Key

Raw access, pay per token. Full control over model and spend.

OAuth / BYO subscription

Multi-provider failover

Define a backup provider in your policy. If Claude goes down, automatically reroute to GPT-4o — or vice versa. Zero code changes, zero downtime. Your agents keep running.

Anthropic

live

Claude Opus, Sonnet, Haiku. Native Claude Code integration.

API key OAuth (Pro/Max)

OpenAI / Codex

live

GPT-4o, o3, o4-mini, Codex. Full model routing support.

API key OAuth (ChatGPT Plus)

Cursor

coming soon

Route Cursor agents through your gateway. Policies, audit, and approvals apply.

OAuth (Cursor Pro/Business)

Google Gemini

coming soon

Gemini 2.0, 1.5 Pro. Google AI Studio and Workspace integration.

API key OAuth (Google)

Mistral

coming soon

Mistral Large, Codestral. European sovereignty, fast inference.

API key

Meta Llama

coming soon

Llama 3.x via self-hosted or Groq/Together. Open weights, full control.

API key self-hosted

// missing your provider? request it →

features

Enterprise-grade controls. Zero infra to manage.

We run the gateway. You define the policies. Your agents stay controlled.

Policy-Based RBAC

Define access in YAML — which AWS actions, git repos, and tools each agent can touch. We enforce it on every request. Deny by default.

Full Audit Trail

Every tool call, approval, and API request logged automatically. Query in the dashboard, stream to your S3, replay any session.

Human-in-the-Loop Approvals

Risky operations gate on your approval. Get notified via Telegram, Slack, or the dashboard. One tap to approve or deny.

Multi-Agent Orchestration

A2A messaging between agents, live session tracking, and a real-time dashboard showing every running agent and its current task.

Cost Controls

Per-project token budgets, rate limiting, and spend alerts. Know exactly what each agent costs before the bill arrives.

Managed Infrastructure

No Postgres to configure, no Redis to tune, no gateway to patch. We handle uptime, backups, and upgrades. You get a URL and a dashboard.

how-it-works

From signup to controlled agents in one afternoon.

Sign up and create your workspace

We provision your managed gateway endpoint — a dedicated URL at you.sentrely.io. No infra setup, no cloud account required on your end.

Define your access policies

Write a project.yml declaring which AWS actions, git repos, and tools each agent is allowed to use. Push it to your workspace. We enforce it.

Point your agents at the gateway

Set GATEWAY_URL in your agent's environment. Every API call, git push, and tool invocation now routes through your control plane.

Monitor, approve, and audit

Watch live sessions from your dashboard. Approve risky ops from Telegram or Slack. Query the full audit log. Export to your own S3 bucket at any time.

architecture

// your agents (anywhere) ┌──────────────┐ ┌──────────────┐ │ claude-agent │ │ claude-agent │ │ (project A) │ │ (project B) │ └──────┬───────┘ └──────┬───────┘ │ │ └─────────┬────────┘ ▼ // your managed gateway (we run this) ┌────────────────────────────────┐ │ you.sentrely.io │ │ policy check → audit log │ │ rate limit → approval gate │ │ HA postgres · Redis · TLS │ └──┬──────────┬──────────┬───────┘ │ │ │ ▼ ▼ ▼ AWS API git push custom tools (scoped) (audited) (sandboxed) // you get dashboard · audit log · approvals

use-cases

Built for every team shipping AI agents.

// solo-dev

Solo Developers

No infra to manage. Connect your AI agents to a fully managed gateway in minutes. Enterprise-grade controls without the enterprise overhead.

→ Policy-scoped AWS access
→ Full audit log, zero config
→ Telegram approval for risky ops

// teams

Engineering Teams

One managed gateway for your whole team. Per-project isolation, shared audit dashboard, agent-to-agent messaging — no Kubernetes required.

→ Multi-project isolation
→ Agent-to-agent messaging
→ Shared audit dashboard

// enterprise

Enterprise

Private deployment in your own VPC. Cross-account IAM, SSO, dedicated support, and a compliance-ready audit trail. We manage the infra inside your perimeter.

→ Private VPC deployment
→ SSO + team management
→ Dedicated SLA + support

pricing

Simple pricing. No credit games.

Every plan includes the full platform. You pay for agents, not features.

// every plan includes

✓ Talk to your agents via Slack or Telegram

✓ Immutable audit trail

✓ Per-agent RBAC policies

✓ Human-in-the-loop approvals

✓ 100+ integrations (AWS, GitHub, Stripe, GHL, Meta + more)

✓ Web dashboard + activity monitor

✓ Bring your own API key (Anthropic, OpenAI, more)

// free trial

14 days free

✓1 agent, full features
✓100 minutes of agent runtime
✓Cancel anytime

Start 7-Day Free Trial

// team

$199/mo

For small teams running real agents

✓10 agents
✓2 seats included
✓Unlimited agent runtime
✓2GB RAM per agent
✓20GB workspace per agent
✓90-day audit retention

Start 7-Day Free Trial

Your agents are live.
Are they controlled?

Speed without controls is reckless. Get a managed control plane and know exactly what your agents are doing, to what, and why.

Book a Demo Instead

The managed control plane for your AI agents.

Watch Sentrely govern an agent — live

Bring your own model. Provider-agnostic by design.

Anthropic

OpenAI / Codex

Cursor

Google Gemini

Mistral

Meta Llama

Enterprise-grade controls. Zero infra to manage.

Policy-Based RBAC

Full Audit Trail

Human-in-the-Loop Approvals

Multi-Agent Orchestration

Cost Controls

Managed Infrastructure

From signup to controlled agents in one afternoon.

Sign up and create your workspace

Define your access policies

Point your agents at the gateway

Monitor, approve, and audit

Built for every team shipping AI agents.

Solo Developers

Engineering Teams

Enterprise

Simple pricing. No credit games.

Your agents are live. Are they controlled?

AI agent stories, every 2 weeks

The managed control plane
for your AI agents.

Your agents are live.
Are they controlled?