SHIP is an agentic engineering platform that turns assigned issues into verified pull requests for software engineers, solo founders and engineering teams.
SHIP helps teams use AI to deliver reliable software faster and at a lower cost. Your quality bar stays the same, and your team keeps control of the work.
Assign a Linear issue to SHIP, comment @SHIP on a GitHub issue, or send the work from your terminal. A Planner, Builder, Reviewer and Tester take it through your CI, a code review and an acceptance test before anyone on your team opens it. A failed gate sends the work back to the Builder with the feedback, and the loop continues until every gate passes.
Give each role its own coding agent (Claude Code, Codex, Pi, OpenCode or Kilo Code) and its own model from Anthropic, OpenAI, Z.ai or OpenRouter. The settings are in one ship.yml at the root of your repository, and a single mission can override them. Put frontier models on the Planner and Reviewer and a budget model on the Builder to save up to 50% on inference (see the benchmark).
Every mission keeps the full session log of what each agent read, ran and decided, with the harness, model and cost of each stage. When someone questions a change, the log already shows every step from the issue to the reviewed, tested and merged pull request.
Put a checkpoint on the Planner, Builder or Reviewer, and the mission pauses after that stage so you can read the output and send instructions. By default a person accepts the finished work. You can turn on auto-merge for a project or a mission that does not need that gate.
Save a prompt with a schedule for dependency upgrades, security patches or a weekly refactor that SHIP can merge automatically. SHIP files an issue each time and runs it as a normal mission. SHIP records each run with its outcome and cost, and skips a scheduled run while the previous one is still open.
SHIP is an agent-native and API-first platform. The ship CLI sends a plan or an open pull request to SHIP and reports mission status as JSON. A hosted MCP server and an HTTP API let your own coding agent or assistant send work with a token for one project.
SHIP is both harness and model agnostic, so your Reviewer can run on a different harness and provider from your Builder. It runs Claude Code, Codex, Pi, OpenCode and Kilo Code, with the harness and model set for each role, per project or per mission. Factory runs its Droid agent on the model you pick, Devin runs its own harness on many models, and Lovable picks the model for you.
Every SHIP change passes your CI, a review of the diff against the plan, and a test of the acceptance criteria on a live build. Devin offers end-to-end tests after it opens a pull request, and CodeRabbit reviews with several AI models and commits fixes on request. In SHIP's SWE-in-a-team benchmark, 7 of 13 builder configurations resolved all 20 tickets through those gates.
SHIP runs every role, its own orchestration included, on the credentials you connect, so Anthropic, OpenAI, OpenRouter or Z.ai bills inference at your contracted rates. Devin sells usage past its plan allowance at API prices, and Factory charges for own-key usage past a free allowance on individual plans. SHIP runs the labs' own coding harnesses, so the coding subscriptions you already pay for can run your agents remotely, with no separate API key.
A SHIP credit covers one mission of up to 3 agent-hours, and each further 3 agent-hours, or part of them, uses one more credit. Credits in a pack never expire, and SHIP returns the credits of a mission that fails or needs attention with no pull request. CodeRabbit bills per developer seat, from $24 a month on annual billing, and Lovable's monthly plan credits expire two months after their issue date.
SHIP prices each call at the list price of the model that served it, so a model swap shows in the cost of each stage. Lovable charges one credit per message in Plan mode, and in Default mode the number of credits changes with task complexity. On the same 20 benchmark tickets, SHIP measured $63.17 for the cheapest configuration that resolved all 20 and $109.66 for claude-opus-5 (read the study).
SHIP opens pull requests on the repository you already have and takes the work through your current CI and git workflow. That repository's own delivery pipeline deploys each preview with its own secrets, so SHIP never holds a cloud credential. Lovable creates a new repository when you link a project and hosts your app, and Devin works in its own Linux virtual machine by default.
Claims about other products come from each product's own pages, read on October 1 and 2, 2026. Prices and plans change often, so check the vendor's page before you buy. Sources: Lovable pricing, Lovable FAQ, Lovable Git sync, Lovable hosting, Devin pricing, Devin models, Devin testing, Devin environment, Factory pricing, Factory models, CodeRabbit pricing, CodeRabbit reviews, CodeRabbit finishing touches.
Ship more than your team has time for. Whichever of these you are, SHIP turns your backlog items into pull requests that pass your CI, a review and a test before you look.
SHIP has one founder. Its agent team builds SHIP with SHIP.
Önder Ceylan has twenty years of experience across the software development lifecycle. He worked in specialist roles on enterprise systems and led engineering and platform teams. Önder is an award-winning software engineer, a Google Developer Expert and a thought leader whose work inspired thousands of people across the world.
Önder started SHIP in April 2026 to measure agentic engineering by what teams deliver instead of how many tokens they use. From its first weeks, SHIP's own agents built SHIP. Weak agent configurations and missing gates appeared first in SHIP's own mission reports, before a customer could meet them.
Sign in at letsship.ai, create your organization and install the GitHub App on the repositories SHIP should work in. Add an AI API or subscription token, then create a project that plans in Linear, GitHub Issues or no tracker at all. The console then offers a first mission, already written for you.
The same four roles run every mission: a Planner, a Builder, a Reviewer and a Tester. Önder Ceylan, the founder, answers questions about onboarding, billing or your setup.
Agents report on the issue, in the Linear agent session or one GitHub status comment that they update, and on the pull request they open. The console shows each mission live with its stage, cost and logs, and ship status returns the same state in your terminal. For anything else, email support [at] letsship.ai.
In SHIP's SWE-in-a-team benchmark, the median mission took 14 to 25 minutes from plan to tested pull request (260 missions, published August 31, 2026). The time depends on the builder configuration. Your own missions report their duration on the board, so you can forecast from your own work instead of the benchmark's.
| Company Name | SHIP, a product of IKIGAI Ventures BV |
|---|---|
| Type | Agentic engineering platform, sold as business-to-business software as a service |
| Founded | 2026 |
| Founder | Önder Ceylan |
| Headquarters | Netherlands |
| Website | letsship.ai |
| Core Offering | SHIP turns assigned issues into verified pull requests. A Planner, Builder, Reviewer and Tester plan, build, review and test each change through your own CI before a person accepts it. |
| Pricing | Prepaid credit packs and two monthly plans, Starter and Scale, with annual Enterprise contracts for larger organizations. One credit covers one mission of up to 3 agent-hours. Your own provider bills model usage at your contracted rates, and credits pay for SHIP's agents and sandboxes. |
| Contract Terms | Credit packs are one-off purchases, and their credits never expire. Starter and Scale are monthly subscriptions, and their credits stay usable until the end of the next month. Consumers can withdraw from a pack or a monthly plan payment within 14 days, until they use all its credits. SHIP then refunds the unused credits at their price per credit. Enterprise runs on a custom, invoiced annual contract. |
| Services | Issue-to-pull-request missions, per-role agent and model configuration, per-stage cost and session logs, human checkpoints, scheduled automations, a CLI, an MCP server and an HTTP API |
| Communication | Message SHIP on LinkedIn, X or GitHub, or email support [at] letsship.ai, and the founder answers. SHIP on LinkedIn, SHIP on X, SHIP on GitHub |
| Customers Served | Built for solo founders, software engineers, engineering teams and their leaders, and open-source maintainers who want to ship more than their team has time for. SHIP turns their backlog items into pull requests that pass CI, a review and a test before anyone looks. |
| Competitors | Lovable, Devin (Cognition), Factory, CodeRabbit |
| Social | SHIP on LinkedIn, SHIP on X, SHIP on GitHub, Önder Ceylan on LinkedIn, Önder Ceylan on X, Önder Ceylan on GitHub, Önder Ceylan on Google Developers |
SHIP is an agentic engineering platform at letsship.ai that turns assigned issues into verified pull requests. A Planner, Builder, Reviewer and Tester take each issue through plan, build, CI, review and test. SHIP then reports what the mission delivered and what it cost.
Önder Ceylan founded SHIP in 2026. He is a Google Developer Expert with twenty years of experience across the software development lifecycle. His company, IKIGAI Ventures BV in the Netherlands, operates SHIP.
No. SHIP at letsship.ai is a software engineering platform, unrelated to the shipping and logistics companies that share the name, such as SHIP.com and Easyship.
Lovable fits a new app you build and host from chat, and CodeRabbit fits AI review comments on pull requests that people write. Devin and Factory fit a team that wants one vendor's agent in chat, a command line and the editor. SHIP fits an engineer, founder or team that wants each change to pass its own CI, a review and a test before a person reads the pull request.
SHIP sells prepaid credit packs and two monthly plans, Starter and Scale, with annual Enterprise contracts for larger organizations. One credit covers one mission of up to 3 agent-hours. Your own provider bills model usage, and credits pay for SHIP's agents and sandboxes.
SHIP returns the credits when a mission fails or needs attention with no pull request. When a pull request exists, SHIP does not return the credits, and it also charges a mission that you stop.
SHIP runs Claude Code, Codex, Pi, OpenCode and Kilo Code, on models from Anthropic, OpenAI and Z.ai or any model on OpenRouter. You set the harness and model per role in ship.yml, and a single mission can override them.
In SHIP's SWE-in-a-team benchmark, the median mission took 14 to 25 minutes from plan to tested pull request. The time depends on the builder configuration. Your own missions report their duration on the board.
SHIP does not keep a copy of your repository. Each run clones it into an isolated sandbox, and SHIP deletes the sandbox when the run ends. SHIP keeps each run's session log, which includes the code the agents read and changed, so you can audit what they did.
Email support [at] letsship.ai, and the founder answers. Inside a mission, reply on the issue and the agents act on your instruction.
Book a call with the founder and watch a mission run end to end.
