🚀 v1.26.0 — Context & observability suite, autonomous Sentry triage, Phoenix LLM tracing, agent-governance controls, Git Sync · What's new

Ship AI Agents From Idea to Production

Agentic mission control: create AI agents with specific roles and goals, connect them into crews and workflows, and deploy with approval gates and budget controls baked in. Operate it all from the dashboard — or let your own agents drive it over MCP.

For humans Sign Up Free

For agents Connect via MCP

See How It Works Star on GitHub

FleetQ Dashboard

Active Agents

+3 today

Experiments

8 running

Budget Used

62%

Content Researcher

Claude Sonnet · 2m ago

Running

Code Reviewer

GPT-4o · Awaiting approval

Pending

Data Analyst

Gemini Pro · Completed

Done

Want to look around first? Try the live demo — no sign-up needed.

60+

Pre-built Integrations

20+

Signal Channels

AI Providers

Workflow Node Types

Build Agents. Control Costs. Scale Confidently.

From a single agent to a coordinated fleet — design, govern, and scale your AI workflows.

Build

Create AI agents with defined roles and goals. Equip them with skills, connect external tools, and choose your preferred LLM provider.

Multi-agent crews & coordination
Visual workflow builder with 26 node types
Skill & tool marketplace
GitHub, Slack, Notion, Linear & more — one click to enable
20+ signal channels with media pipeline
Cross-channel contact identity
User Browser Relay & Puppeteer/Playwright MCP
Plugin system — add MCP tools, signals & UI panels via Composer
Self-healing browser harness — agents add their own helpers mid-task
Auto-MCP — connectors expose themselves as MCP tools, no boilerplate

Control

Keep humans in the loop with approval gates, budget enforcement, and a full audit trail for every action.

Human-in-the-loop approval gates
Per-tier risk policy — auto / ask / reject for read · write · destructive
Budget caps & real-time cost tracking
AI Safety Guardrails (PII, toxicity, budget)
Full audit trail with decision reasoning
Agent risk profiles & auto-disable
Customer-managed encryption keys (BYOKMS)
Real-time AI streaming in the assistant

Scale

Go from one agent to an entire fleet. Run structured experiments, iterate on results, and track every metric.

20-state experiment pipeline with auto-retry
Parallel and sequential workflow execution
AI spend forecasting & semantic caching
7 AI providers with automatic cloud failover
Event-driven trigger rules & scheduling
AI assistant with workflow delegation
Live team graph & real-time WebSocket events
New Self-improving agents — memory consolidates lessons and patterns between sessions
New Workflow YAML Git Sync — workflows as source of truth in your repo

From Idea to Production in Three Steps

Go from idea to a running AI workflow in three steps.

Design

Create AI agents with specific roles and goals. Attach skills, tools, and credentials. Choose your LLM providers.

Build

Assemble multi-agent crews or design workflows with the visual editor. Add conditional logic, human tasks, and parallel branches.

Deploy

Run experiments with human approval gates, budget controls, and real-time metrics. Iterate on results and scale what works.

Try It in 5 Minutes

One Platform, Five Plans

Every plan includes the full platform. Pay only when you need more capacity.

Every plan includes: BYOK multi-provider support, visual workflow builder, audit trail, REST API, and MCP server.

Free forever

Free

For individuals and side projects. No credit card required — start building in minutes.

0 € /month

Start Free

What's included

25 agents
300 runs/month
2 team members
25 workflows
7-day audit retention
Full platform access (BYOK, MCP, API)

Need more? Scale up

Starter

For small teams getting started

19 € /month

75 agents
1,000 runs/month
3 team members
100 workflows
30-day audit retention
Advanced approval workflows
Private marketplace publishing

Get Started

Pro

For teams with production workloads

59 € /month

Unlimited agents
10,000 runs/month
10 team members
Unlimited workflows
90-day audit retention
Advanced approval workflows
Public marketplace publishing
Priority support

Get Started

Business

For scaling teams with advanced needs

220 € /month

Unlimited agents
50,000 runs/month
25 team members
Unlimited workflows
1-year audit retention
Advanced approval workflows
Public marketplace publishing
Custom branding
Priority support

Get Started

Enterprise

For organizations with custom requirements

Custom

Unlimited agents
Unlimited runs/month
Unlimited team members
Unlimited workflows
10-year audit retention
Advanced approval workflows
Public marketplace publishing
Custom branding
Priority support

Contact Sales

Frequently Asked Questions

FleetQ is an AI Agent Mission Control Platform. You create AI agents with specific roles and goals, assemble them into multi-agent crews, and orchestrate their work through visual workflows. Every run includes human-in-the-loop approval gates and budget controls, so you stay in charge.

Yes. FleetQ is fully open source under the MIT license. Self-host it on your own infrastructure with complete control over your data, models, and configuration.

FleetQ supports 10+ cloud providers — Anthropic (Claude), OpenAI (GPT-4o), Google (Gemini), Groq, OpenRouter, Mistral, DeepSeek, xAI (Grok), Perplexity, and Fireworks — with automatic failover. You bring your own API keys. Local CLI agents (Claude Code, Codex, Gemini CLI, Kiro, Aider, Amp, OpenCode, Ollama, and more) are also supported at zero platform cost.

Yes. FleetQ ships as a Docker stack with PHP 8.4, PostgreSQL 17, and Redis 7. Run the install wizard and it walks you through database setup, admin account creation, AI provider keys, and default agent configuration in under five minutes.

Every LLM call has a credit cost calculated from token usage. You set budget caps at the global and per-experiment level. The platform alerts you at 80% usage and automatically pauses operations when the budget is exhausted — no surprise bills.

FleetQ includes a Model Context Protocol (MCP) server with 670+ tools across 60+ domains. External AI agents — Claude Code, Codex, Gemini CLI, Kiro, Amp, Cursor, and more — can manage your FleetQ instance directly: creating agents, triggering workflows, checking budgets, and browsing the marketplace.

All API keys and secrets are encrypted at rest using dedicated per-team keys with XSalsa20-Poly1305 authenticated encryption. Pro and Enterprise teams can connect their own AWS KMS, GCP Cloud KMS, or Azure Key Vault — revoking KMS access in your cloud provider immediately blocks all credential decryption. The platform also enforces rate limiting, budget caps, target blacklists, a complete audit trail, and role-based access control with four permission levels.

Every team's credentials are encrypted with a dedicated key using XSalsa20-Poly1305 authenticated encryption. Pro and Enterprise teams can bring their own AWS KMS, GCP Cloud KMS, or Azure Key Vault to wrap the encryption key — revoking KMS access in your cloud provider immediately blocks all credential decryption.

FleetQ supports 20+ inbound signal channels: Telegram, WhatsApp, Discord, Slack, Signal Protocol, Matrix, GitHub webhooks and issues, Jira, Linear, Sentry, PagerDuty, Datadog, RSS, IMAP, HTTP Monitor, Calendar, API Polling, manual signals, and generic webhooks. Inbound messages — including photos, audio, video, and documents — are automatically processed and routed to agents, crews, or trigger rules.

FleetQ automatically resolves a single contact record across all inbound channels. If the same person messages you on Telegram, WhatsApp, and Discord, they appear as one unified contact with a full interaction history. You can manually merge or unlink identities, and agents receive the full contact context when processing signals.

Yes. FleetQ has a built-in User Browser Relay: generate a relay token in Settings, run the open-source relay agent binary on your machine alongside BrowserMCP, and FleetQ agents can control your real Chrome browser — including authenticated sessions and cookies. Only 13 specific browser tools are permitted and all commands are routed through the platform and logged. For headless scraping without a local binary, Puppeteer and Playwright are available as pre-built MCP integrations.

Yes. FleetQ has a first-class plugin system. On self-hosted installations, plugins are standard Composer packages — install with composer require and enable in Settings → Plugins. On the managed cloud, the platform pre-installs a curated set of plugins and each team can enable or disable them from Settings → Plugins without touching the server.

Event-driven trigger rules let you automatically start experiments and projects when inbound signals match defined conditions. You configure rule conditions (equals, contains, threshold), map signal fields to project inputs, and set cooldown and concurrency limits — no manual intervention needed.

The built-in AI assistant is context-aware — it understands the page you are on and gives you direct access to all platform capabilities through 675+ tools. It streams responses in real time and can delegate work to your projects and workflows, then notify you via the app or Telegram when the results are ready. Risky operations (writes, integrations, git pushes) can be gated behind a per-tier approval policy from Team Settings.

All paid plans are billed monthly in EUR via Stripe. You can upgrade, downgrade, or cancel anytime. The free plan requires no credit card.

Yes. You can change your plan at any time from the billing page. Upgrades take effect immediately and are prorated. Downgrades take effect at the next billing cycle.

You will receive alerts at 80% and 100% of your usage limits. When a limit is reached, new operations in that category are paused until the next billing cycle or until you upgrade.

Ready to Put Your AI Agents to Work?

Free to sign up. Pay only for what you use — starter pack from €8.