CrewAI
by CrewAI, Inc.
Role-based multi-agent orchestration, from open-source Python to a governed enterprise runtime
CrewAI is an MIT-licensed Python framework for orchestrating teams of AI agents, plus AMP, a commercial platform that deploys, monitors and governs those agents in production. It pairs autonomous role-based Crews with event-driven Flows, so teams can mix loose agent collaboration and deterministic control in one system.
CrewAI is a Python framework and commercial platform for building multi-agent systems, created by Joao Moura and first published to PyPI as version 0.1.3 on 4 December 2023. The open-source core is MIT-licensed and describes itself as 'a lean, fast Python framework built specifically for orchestrating autonomous AI agents and production-ready agentic workflows'; it is explicitly standalone, with 'its own primitives for agents, tasks, crews, flows, tools, and orchestration' and no LangChain dependency. The framework exposes two complementary abstractions. Crews give agents specialised roles, goals and tools and let them collaborate autonomously, which suits open-ended problems. Flows are event-driven workflows offering precise execution control, state management, conditional branching and native support for invoking Crews, which suits paths that must be deterministic; the vendor's position is that production systems combine both. On top sits CrewAI AMP, the commercial runtime, which adds managed crew deployment, a REST API for deployed crews, execution traces and logs, a shareable tool repository, webhook event streaming and Crew Studio, a no-code and low-code builder. Deployment into AMP happens through GitHub integration, Crew Studio or the CLI. The company raised $18 million in October 2024 across a seed round led by boldstart ventures and a Series A led by Insight Partners, with Craft Ventures, Earl Grey Capital, Andrew Ng and HubSpot co-founder Dharmesh Shah participating; more than 150 enterprises signed up during the Enterprise beta. CrewAI reports 65% of the Fortune 500 as users, 450 million-plus agentic workflows run monthly, 4,000 weekly sign-ups and named customers including DocuSign, PepsiCo, IBM and Experian. The repository carries roughly 56.9k stars and 8.1k forks.
The enterprise platform team that already has business units requesting agents and needs one governed build-and-runtime layer, with a no-code studio for the business side and Python for engineers.
Agents move from a notebook prototype to a deployed, traced, access-controlled service with SOC 2 Type 2 controls behind them, without the team building the deployment and observability layer themselves.
At a Glance
- Category
- AI Agents & Orchestration
- Pricing
- Freemium, Contact for pricing, Free
- Target Market
- CIOs, CTOs, Enterprise Developers, Automation Leads
- Deployment
- Open-source, Self-hosted, Cloud-first, Hybrid
- Founded
- 2023
- Customers
- Vendor reports 65% of the Fortune 500 as users and 4,000+ weekly sign-ups; 150+ enterprises joined the Enterprise beta
Key Features
- ✓Crews — role-based agent teams
Agents are given specialised roles, goals and tools and collaborate autonomously, which suits problems where the path is not known upfront.
- ✓Flows — event-driven orchestration
Deterministic workflows with state management and conditional branching that can invoke Crews, giving precise control where autonomy is unacceptable.
- ✓Standalone Python framework
Ships its own agent, task, crew, flow and tool primitives with no LangChain dependency, keeping the dependency surface and upgrade risk small.
- ✓Crew Studio no-code builder
A visual editor with an AI copilot lets business teams assemble agent workflows without writing Python, under platform-team governance.
- ✓Managed deployment and REST API
Crews deploy to managed infrastructure from GitHub, Crew Studio or the CLI and are then callable as REST endpoints from existing systems.
- ✓Observability and webhook streaming
Detailed execution traces and logs plus real-time webhook events let teams debug agent runs and pipe activity into existing monitoring.
- ✓Enterprise governance controls
SSO, role-based access control, immutable audit trails and runtime PII redaction hooks injected at every LLM and tool call.
- ✓Tool repository
Teams publish and install shared tools across crews, so integrations are written once instead of re-implemented per project.
Capabilities
Use Cases
- •Lead enrichment and routing at scale
Gelato reports over 3,000 leads enriched per month and DocuSign 75% faster first contact with inbound leads using deployed crews.
- •Customer support response automation
Piracanjuba reports 95% response accuracy for customer support, with agents drafting and routing replies against internal knowledge.
- •Quality assurance and review workloads
Konecta reports a 96% reduction in QA time by handing structured review passes to agent crews rather than analysts.
- •Internal research and report generation
A crew of researcher, analyst and writer agents gathers sources, reconciles findings and produces a drafted report for human review.
- •Accelerating internal development work
General Assembly reports a 90% reduction in development time by assembling agent workflows instead of hand-building each integration.
Ideal For
Best For
- ✓Orchestrating teams of specialised agents with distinct roles, goals and tools on open-ended tasks
- ✓Mixing autonomous collaboration and deterministic branching in one system via Crews plus Flows
- ✓Deploying agent workflows to managed infrastructure with execution traces, logs and webhook streaming
- ✓Letting non-engineers assemble agent workflows in Crew Studio under platform-team governance
- ✓Standardising and sharing internal agent tools across teams through the tool repository
Not Ideal For
- ✗Organisations that cannot accept the SaaS terms of service, which grant CrewAI rights to use customer data to train and develop models with no public opt-out — the open-source self-hosted path is the alternative
- ✗Teams relying on smaller open-weight local models, since practitioners report most OSS Ollama models struggle with the tool-calling the framework depends on
- ✗Non-Python engineering organisations, as the framework and its primitives are Python-only
- ✗Buyers who need EU data residency or an ISO 27001 certificate, given the managed platform runs on AWS US and neither GDPR DPA nor ISO 27001 is publicly confirmed
Integrations
Deployment
Market & Ratings
Vendor reports 65% of the Fortune 500 as users and 4,000+ weekly sign-ups; 150+ enterprises joined the Enterprise beta
Market Analysis
Pros
- ✓Role-based Crews are the most approachable mental model in the category — engineers get a working multi-agent prototype quickly
- ✓Flows cover the deterministic half of the problem, so teams are not forced to choose between autonomy and control
- ✓MIT licence with no LangChain dependency makes self-hosting and air-gapped deployment genuinely viable
- ✓Real enterprise governance shipped: SOC 2 Type 2, SSO, RBAC, immutable audit trails and PII redaction hooks at every LLM and tool call
- ✓Very large community — roughly 56.9k GitHub stars, 8.1k forks and 100,000-plus developers through the free certification courses
Cons
- ✗The SaaS terms of service grant CrewAI rights to use customer data 'to improve and develop products and services, including by training and developing models', with no public opt-out — a hard blocker for many procurement teams
- ✗Telemetry has been a live complaint on Hacker News, with a thread titled 'CrewAI collects prompts via telemetry, can't be disabled'
- ✗Practitioners report that 'most OSS Ollama models struggle' with the tool-calling the framework requires, effectively tying good results to frontier hosted models
- ✗Adoption sentiment is contested — a Hacker News thread asks outright 'Does anyone use CrewAI or LangChain anymore?' as newer frameworks arrive
- ✗No published middle pricing tier: the free plan stops at 50 workflow executions a month, so any real usage means an enterprise sales cycle
- ✗The managed platform runs on AWS US with no confirmed GDPR DPA or ISO 27001 certificate, which limits EU deployments
Pricing
Open Source Framework
$0
- ✓MIT-licensed Python framework
- ✓Crews and Flows
- ✓Self-hosted, including air-gapped
- ✓Community support
Basic (AMP)
$0
- ✓50 workflow executions per month
- ✓Visual editor and AI copilot
- ✓GitHub integration
- ✓Tracing and community support
Enterprise (AMP)
Contact for pricing
- ✓SSO, RBAC, workload identity and PII redaction
- ✓Flexible deployment options
- ✓45-day production onboarding program
- ✓Dedicated support
The framework itself is MIT-licensed and free to self-host indefinitely. On the commercial AMP platform only two tiers are published: a Basic plan at no cost capped at 50 workflow executions a month, and Enterprise at undisclosed custom pricing — there is no published middle tier, so any real production volume means a sales conversation. Model inference is billed separately by whichever provider you point the agents at, and the 50-execution ceiling makes the free tier a trial rather than a working plan.
Security & Compliance
Connect
Sources
This page was written from 8 sources, 6 on domains other than crewai.com.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Sema4.ai
Enterprise AI agents for knowledge work, running natively in your Snowflake or cloud account
Sapiom
Agent infrastructure that routes, runs and meters AI agents in production
Itential FlowAI
Governed AI agents for network and infrastructure operations, with deterministic execution and full audit trails
OpenAI Presence
Deploy production-grade AI voice and chat agents with enterprise policies, guardrails and evals
Mentioned In
Perplexity Bets $20B That Multi-Model Orchestration Beats Single-Vendor Lock-In
Perplexity Bets $20B That Multi-Model Orchestration Beats Single-Vendor Lock-In. For CFOs and finance leaders: cost implications, budget planning, and ROI be...
March 12, 2026Enterprise AIThe $670K Gap: Why 78% of Enterprise AI Pilots Die Before Production—And How Governance Fixes It
Multi-agent AI systems promise 5x-10x returns, but 88% fail before launch. The missing ingredient isn't smarter models—it's operational infrastructure: orchestration, monitoring, and governance.
March 27, 2026Enterprise AIWhy 79% Pilot AI Agents But 11% Deploy: Nexus $4.3M Bet
Nexus raises $4.3M to close the 68-point gap between AI agent adoption and production. Orange Group saw 50% conversion lift in 4 weeks.
March 31, 2026