Crewship
Crewship is a developer-first cloud platform that takes CrewAI, LangGraph and LangGraph.js agents from local code to production with a single command, adding managed infrastructure, a REST API, real-time streaming, versioning and execution traces.
What is Crewship?
Crewship is a developer-first cloud platform for deploying and running AI agent workflows. It does not help you design an agent; it takes one you have already written and puts it into production. Three frameworks are supported today — CrewAI and LangGraph in Python, and LangGraph.js in TypeScript or JavaScript — with the same deployment and invocation experience across all of them. Six more, among them Agno, AutoGen, Pydantic AI, smolagents, OpenAI Agents and Mastra, are advertised as coming soon.
The workflow is deliberately short. You write your agent locally, add a crewship.toml file and run crewship deploy. The CLI detects the project, packages it, builds a container image and returns a deployment identifier and a version number. No Docker or Kubernetes knowledge is required, and the vendor claims a first deployment in under five minutes.
What comes back is a production API endpoint. Two invocation modes share it: the Runs API handles stateless, fire-and-forget executions such as research, analysis or content generation, while the Threads API keeps persistent state across messages for chatbots and support agents. Requests are authenticated with a bearer token. Execution can be watched live through Server-Sent Events or reported through outgoing webhooks signed with HMAC-SHA256.
Around that core sits the operational tooling that separates a platform from a script. Every deployment is versioned and can be rolled back instantly, keeping staging and production apart. Execution traces record each agent action, LLM call and tool use, alongside token consumption and cost. Files an agent writes to artifacts/ are collected automatically and retrievable through the API. Structured data tables can be read and written by agents, with filtering, sorting and pagination. Secrets are encrypted at rest and in transit and injected at runtime into isolated environments, and workloads scale from zero to thousands of concurrent runs.
The service is operated by Halasi AI Lab, based in Emmen, Switzerland, and governed by Swiss law. It is a young offering: the domain was registered in January 2026 and the site was first archived on 28 January 2026.
What it does
- Deploy a CrewAI, LangGraph or LangGraph.js agent to production with one command
- Expose every deployment as a REST API, stateless through Runs or stateful through Threads
- Stream agent events and logs live over Server-Sent Events
- Version each deployment and roll back to any earlier one instantly
- Trigger runs from incoming webhooks or on a cron schedule
- Collect the files agents generate and download them through the API
- Connect a deployed agent to Slack through @mentions and slash commands
When to use Crewship / When not to
A quick filter to help you decide if Crewship is the right fit.
When to use Crewship
- Engineering teams with a working CrewAI or LangGraph agent that now needs a production endpoint, monitoring and rollbacks
- Solo developers and small startups who want a permanent free tier and flat pricing with no per-seat fees
- Teams building chatbots or support agents that need persistent multi-turn state through the Threads API
- Developers who want an agent reachable from Slack via @mentions and slash commands without building the integration themselves
- Automation-minded teams triggering agent runs from webhooks or cron schedules and collecting the files those runs produce
When not to use Crewship
- Non-developers: there is no visual builder and no no-code mode, everything goes through a CLI and a code project
- Teams committed to AutoGen, Agno, Pydantic AI, smolagents, OpenAI Agents or Mastra, all still marked as coming soon
- Organisations that require self-hosting, which the vendor states is not currently supported
- Regulated or security-sensitive buyers needing a signed DPA, a SOC 2 or ISO 27001 certification, or a guaranteed hosting region
- Anyone needing an evaluation framework, prompt management or a visual graph debugger, which the vendor's own comparison lists as absent
How to use Crewship
A typical end-to-end flow, from setup to results.
- Create an account on the Crewship console
- Have an agent project ready in CrewAI, LangGraph or LangGraph.js, or clone one of the quickstart repositories published on GitHub
- Install the CLI: one command via curl on macOS and Linux; on Windows, download the executable, add it to your user PATH and restart the terminal
- Authenticate the CLI with crewship login
- Add a crewship.toml configuration file to the project, or accept the defaults; several deployments can be declared in a single file
- Register environment variables and secrets with crewship env so credentials are injected at runtime
- Run crewship deploy to package, build and publish the agent, then note the deployment identifier and version returned
- Invoke the agent with crewship invoke, with a POST to the Runs or Threads API, or directly from the console
- Watch events and logs stream live over Server-Sent Events, or register an outgoing webhook to be notified when a run completes
- Review runs, artefacts, traces and schedules in the console, and roll back to an earlier version if a deployment misbehaves
Pros & Cons
Pros
- Going to production takes a single command, with no Docker or Kubernetes to configure
- One workflow covers three frameworks, including a TypeScript route through LangGraph.js
- Flat, readable pricing at USD 25 per month, with no per-seat charge, no uptime fee and no per-execution overage
- The Pro plan includes 500 runs per month against 100 for CrewAI AMP at the same price, according to the vendor's comparison
- A permanent free plan is available without a credit card, and paid runs may last up to three hours
- Observability comes built in: traces, token costs, instant rollback and log retention of up to one year
- Ownership of deployed code and data stays with the customer, with an explicit commitment not to share it with third parties
Cons
- The service is very young: the domain was registered in January 2026 and the blog holds only four posts
- The vendor is hard to verify — Halasi AI Lab appears nowhere outside its own site, with no team page, no social profiles and one email address serving support, legal and privacy alike
- Six of the nine advertised frameworks are still marked coming soon, and self-hosting is not supported
- The Team plan is unreleased, so shared deployments, role-based access and audit logs are unavailable, while SSO and an SLA remain Enterprise-only
- No security certification, no data processing agreement and no Article 27 EU representative are published
- No hosting country or region is guaranteed, only Cloudflare's global infrastructure, with transfers outside the EEA possible
- The free plan is tightly capped at one deployment, 30 runs and six minutes of runtime per month, and there is no trial of the Pro plan
Pricing & Plans
A permanent free plan is available at no cost and without a credit card, limited to one active deployment, thirty runs per month and six minutes of runtime. The lowest paid entry point is the Pro plan at USD 25.00 per month, billed monthly through Stripe, which raises the allowance to ten deployments and five hundred runs per month; each additional deployment is charged at USD 2.50 per month. The Team plan has been announced but is not yet released, and Enterprise pricing is quoted on request.
- 1 active deployment
- 10 versions per month
- 30 runs per month
- 6-minute maximum runtime
- 1 concurrent run
- 7-day log retention
- community support
- 10 deployments included plus USD 2.50 per month for each additional one
- unlimited versions
- 500 runs per month
- 3-hour maximum runtime
- 10 concurrent runs
- artefacts up to 100 MB
- 30-day log retention
- priority email support
- everything in Pro plus team collaboration
- shared deployments
- role-based access and audit logs
- the comparison table lists unlimited deployments
- 1000 runs per month
- 50 concurrent runs
- artefacts up to 500 MB
- 90-day log retention and SSO/SAML
- unlimited deployments and runs
- custom runtime limits
- dedicated infrastructure
- custom artefact sizing
- one-year log retention
- SSO/SAML
- audit logs
- an SLA guarantee
Data, GDPR & hosting
A consolidated view of how Crewship handles your data.
GDPR overview
GDPR is addressed concretely rather than merely claimed. The privacy policy, effective 9 February 2026, sets out legal bases for the EEA, the UK and Switzerland — consent, performance of a contract, legitimate interests and legal obligations — and lists the rights of access, rectification, erasure, restriction, portability, objection and withdrawal of consent, all exercised through a single mailbox. International transfers over Cloudflare's infrastructure are covered by standard contractual clauses approved by the European Commission or by the provider's adherence to recognised frameworks, and complaints may be filed with a local supervisory authority or the Swiss FDPIC. Two gaps deserve mention: no Article 27 EU representative is designated even though the vendor is established outside the Union, and no data processing agreement is published or offered on request.
Who owns the data?
Under the terms of service you keep ownership of everything you deploy: code, configurations and data are defined as User Content, and Halasi AI Lab receives only a limited licence to host, run and store them for the sole purpose of delivering the service. Intellectual property in your code stays with you, while the platform itself remains the vendor's. The privacy policy adds that deployed code, run data and agent outputs are never shared with any third party and are accessible only to you and members of your organisation. The vendor also states that it does not review, monitor or modify your code before or during execution.
Reuse rights
You may reuse everything you deploy and everything your agents produce, without asking permission: ownership never transfers, and the licence granted to Halasi AI Lab is limited to hosting, running and storing your content in order to operate the service. Build artefacts, run logs, outputs and the files collected under artifacts/ are retrievable through the API and remain yours to use as you see fit, commercially or otherwise. The restrictions run the other way: you may not copy, modify, distribute or reverse-engineer the Crewship platform, resell the service without written consent, or circumvent usage quotas and billing controls. Personal data collected about you is processed only for the purposes listed in the privacy policy, and the vendor states that it does not sell personal information.
Data retention & training
Hosting summary
Hosting is delegated to Cloudflare, named in the privacy policy as the provider of hosting and infrastructure for the platform. No country and no region are committed to: the policy speaks only of Cloudflare's global infrastructure and warns that personal information may be transferred to and processed in countries outside your own, including outside the EEA, the United Kingdom and Switzerland. Where such transfers occur, the stated safeguards are standard contractual clauses approved by the European Commission, or the provider's adherence to recognised data protection frameworks. Payments are handled separately by Stripe, which stores the card details; the vendor states that it keeps no card or bank account numbers on its own servers. Halasi AI Lab is established in Emmen, Switzerland, and the terms are governed by Swiss law with jurisdiction in Lucerne, but nothing indicates that data is hosted there. The security measures described are isolated execution environments and secrets encrypted at rest and in transit; no certification such as SOC 2 or ISO 27001 is published, and no data residency option is offered.
Things to keep in mind
Risks and trade-offs to weigh before adopting Crewship.
- The company behind the service is essentially unverifiable from the outside — no registry entry, no named team, no social presence — which is worth weighing before entrusting it with anything critical
- Availability is provided as is and as available, with no SLA below Enterprise, and liability is capped at the fees paid over the preceding twelve months
- Deployed code and data are deleted within thirty days of termination or suspension for a terms violation, so keep your own copies of everything that matters
- Fees are non-refundable and no prorated refund is granted for a partial billing period
- Your agents call third-party APIs such as OpenAI, Anthropic or Serper, and those costs, terms and outages remain entirely your responsibility
- The vendor does not review or monitor the code it runs, so responsibility for what an autonomous agent does — runaway loops and unintended external calls included — sits with you
- Data location is not guaranteed beyond Cloudflare's global infrastructure, and the resolved server IP is an anycast node that says nothing about where data actually rests
Setup & Integrations
Technical difficulty
Moderate, and strictly for technical users. Everything runs through a terminal and a code project; there is no visual builder. Installing the CLI is one command on macOS and Linux, while on Windows it means downloading the executable, editing the user PATH and restarting the terminal. You then authenticate, add a crewship.toml file — the defaults are usable as they are — and deploy. No Docker or Kubernetes knowledge is needed. The real prerequisite is a working CrewAI, LangGraph or LangGraph.js agent and the ability to read Python or TypeScript; quickstart repositories exist for those starting from scratch.
Deployment
Integrations
Supported languages
Behind Crewship
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Crewship.
Frequently asked questions
Which agent frameworks can I deploy on Crewship?
Do I need to manage any infrastructure?
Can I run Crewship on my own servers?
Is there a free plan, and is there a trial of the paid plan?
What does the paid plan cost?
How are secrets and API keys handled?
Who owns the code and data I deploy?
How long are my data and logs kept?
Is there a minimum age to use the service?
Should you pick Crewship?
Crewship occupies a narrow and useful slot: it does not help you build an agent, it takes one you have already written and gives it a production home. For a team with a CrewAI or LangGraph project running locally, the proposition is easy to evaluate — one command, a REST API, live streaming, versioning with instant rollback, execution traces and automatic artefact collection, for USD 25 a month with no per-seat or per-execution charges. On the vendor's own figures that buys five times the monthly runs of CrewAI AMP at the same price, and the permanent free plan makes it testable without a card.
The reservations concern maturity rather than design. The domain was registered in January 2026, the blog holds four posts, and Halasi AI Lab leaves almost no trace outside its own website: no team page, no social profiles, no company registry entry, and one email address serving support, legal and privacy alike. Governance is correspondingly thin for enterprise use — no data processing agreement, no security certification, no Article 27 EU representative despite the vendor being established outside the Union, and no committed hosting country beyond Cloudflare's global infrastructure. Self-hosting is not offered, six of the nine advertised frameworks are unreleased, and the Team plan does not yet exist, which leaves shared deployments and audit logs out of reach below Enterprise.
What the vendor does put in writing deserves credit: you keep ownership of your code and data, and it is not shared with third parties. Crewship is a sensible choice for prototyping and for small-scale production where downtime is survivable. For regulated or critical workloads, verify the company behind it first.
- Choosing a selection results in a full page refresh.
- Opens in a new window.