covey and CrewAI
A framework for programming agents, and a platform for running them.
Last checked: 29 September 2026, against CrewAI's repository, platform documentation and pricing page. If something here is out of date, tell us and we will correct it.
CrewAI is one of the best-known open-source frameworks for agents that work in roles towards goals. You write agents, tasks and crews in Python. AMP, CrewAI's commercial platform, deploys, monitors and scales them.
covey starts one level further along. It does not help you program an agent. It runs agents whose engine already exists, a coding harness such as Claude Code, and gives each of them a workplace, access, rules and a recording. Where CrewAI and covey meet is the question of how agents are operated in a company.
Several of CrewAI's enterprise features are listed only on its pricing page. Rows that rest on those statements are marked as vendor claims.
What it is
| Topic | CrewAI | covey |
|---|---|---|
| Kind of product | An open-source Python framework, plus AMP, a “platform for deploying, monitoring, and scaling your crews and agents in a production environment”.[1][2] | A self-hosted platform that runs agents and treats them like staff: a role, a manager, a backlog, a budget. |
| Defining an agent | Python code: agents, tasks, crews and flows.[1] | Markdown files (SOUL, PLAYBOOKS, ACCESS, HEARTBEAT), versioned and changed by review. |
| Engine | The agent is the program you write with the framework.[1] | An existing coding harness such as Claude Code runs as the engine. covey does not run CrewAI crews. |
| Licence | The framework is MIT. AMP is a commercial product with a free Basic tier and an Enterprise tier.[1][3] | The whole platform is AGPL-3.0; the plugin SDK is MIT. |
Operation and governance
| Topic | CrewAI | covey |
|---|---|---|
| Where it runs | Enterprise: “Deploy on CrewAI cloud, your own VPC, or your own infrastructure”.Vendor claim[3] | On your own hosts, with no contract needed to run it. |
| Security features | Enterprise: SSO (Entra, Okta), RBAC, workload identity, PII redaction, policies.Vendor claim[3] | Login from the first start. Roles for organisation admin, agent owner, security/compliance, auditor and controlling. Access brokered per run. |
| Guard-rails and approvals | Guardrails and human-in-the-loop are listed in both tiers.Vendor claim[3] | Rules per action (allow, require approval, deny), enforced outside the runtime at the broker, the egress proxy and the action layer. |
| Observability | Tracing, with OpenTelemetry.Vendor claim[2][3] | A recording of every run with its tool calls, and token cost per agent and model. |
Also different
- A workplace per agent
- covey starts every agent in its own container with a persistent home directory, started through a runner, also on a single machine. We did not find a public description of how AMP isolates a crew at runtime, so we make no statement about it.
- People in the org chart
- covey puts people and agents in the same org chart, with departments and cost centres, so an agent has a manager and its spend has an owner.
Where CrewAI is ahead
- Developer community, tutorials and name recognition. Agents, tasks and crews are common vocabulary.
- Flexibility: any flow and any tool you can write in Python.
- Enterprise sales, support and compliance commitments.
- A free hosted tier to start without a contract.
Which one fits
CrewAI
Choose CrewAI if you want to build your own agents in Python with full control over the flow, and deploy them on a managed or self-hosted platform.
covey
Choose covey if your agents already exist as a harness such as Claude Code, and the open questions are where they work, what they may do and who can see what they did. It needs no code to describe, hire and limit an agent.
Sources
Checked on September 29, 2026. Statements about CrewAI come from its own public sources: documentation, repository, release notes or website. Anything marked as a vendor claim appears only on marketing or pricing pages.
Other comparisons
Paperclip
The closest relative: open source, self-hosted, agents run like staff. Paperclip runs agents as host processes by default; covey gives each one its own container.
Checked on September 29, 2026Personal and organisational agents (cloud)OpenAI dots
dots are always-on agents that run in OpenAI's cloud on GPT-6 Astra. covey runs an organisation's agents on its own servers, with Claude Code or Codex as the engine.
Checked on September 29, 2026Governance layerMicrosoft Agent 365
Agent 365 registers and governs agents in Microsoft's cloud. covey is a self-hosted platform where agents do their work.
Checked on September 29, 2026SaaS workforceRelevance AI
Relevance AI builds and runs agent teams in its own cloud, without code. covey runs agents on your hosts, each in its own container.
Checked on September 29, 2026