makerOS leads your AI development through fixed quality gates that cannot be talked past. Local, traceable and with no vendor lock-in. With your CLIs and your models.
evaluation is free · runs on your machines · no cloud required
Four patterns emerging in almost every software team right now. Usually all at once.
Every developer uses a different tool. Nobody knows what things are built with, or where the data flows.
Every tool switch starts from zero. Decisions live in chat histories that are gone tomorrow.
One vendor, one pricing model, one dependency. Switching means explaining everything again.
AI code lands in the repo because it looks plausible. Not because it was verified.
The problem is not the AI. The problem is the missing governance.
Not a feature catalog, but six control guarantees. Each one is usable today and verifiable in the product.
The PIV engine leads every story through plan, implement, verify, security and commit. It checks the definition of ready, sets the gates and blocks the commit via a Git hook until security is green. The agent cannot wave itself through.
The security agent checks every story before the commit closes. Always, with zero extra effort. If you want to go one step further, you add a second CLI as an independent review partner: two machines, four eyes, right in the development flow.
A local memory database holds projects, sessions, plans and decisions. Searchable by keyword and by meaning. You switch tools without explaining the project again. State lives in the database, not in a chat history.
You bring your own subscriptions. Claude Code, Codex, Antigravity, Vibe, Cline, OpenCode and Grok run as equal coding CLIs, including automatic MCP wiring on project import. Plus every coding AI on the market, local models too. And you pick per role: build, verify, chat.
Fully local development is possible. Local models via Ollama, voice control that never leaves the machine, no account and no tracking. makerOS contains no AI model of its own. Your policy decides which models run.
The reality check simulates market reactions in a 17-step pipeline: competition, 24 personas, pricing and a verdict with confidence, before weeks are invested. Planning additionally weaves in 175 curated knowledge files, from sources like OWASP, NN/g and GOV.UK.
Not a new process you have to roll out. makerOS runs it: sketch, remember, plan, build, verify, hand over.
One call delivers the last session with its summary, open handoffs and the quality criteria in force. No re-reading and no explaining again.
Human and AI work on the same whiteboard. The CLI draws architectures in one go and attaches sources as notes. It never touches your hand-drawn elements.
Story planning searches the project memory and pulls the right rules from 175 knowledge files: privacy, accessibility, security. And all of it before a single line of code exists.
Plan, implement, verify, security, commit. The PIV engine enforces the order, moves the Kanban card and writes the history.
The security agent checks the story. Only once the security gate is green can the commit close. Optionally, a second CLI reads along as well.
One line of summary in the evening. Tomorrow morning it is right there again, for everyone on the project.
makerOS fully supports agile development. The AI agents are trained on it: they create epics and stories automatically and fill them with tasks and acceptance criteria.
Board, whiteboard and terminal are real screenshots from the app. The reality check is shown as a schematic.
Cards move from planned to done. The engine moves them along, and you always see what is being built right now. Files attach directly to the card.
Human and AI draw on the same surface. The CLI drops complete diagrams onto the board in one go. One click lifts the essence into project memory.
The built-in terminal runs Claude Code, Codex, Antigravity, Vibe, Cline, OpenCode and Grok as equals. Plus every coding AI on the market, local models too. Switching costs one click, not your context.
The pipeline tests an idea against the market in 17 steps: competition, 24 simulated personas, pricing analysis. At the end there is a report with a clear verdict.
makerOS Solo is in productive use. What a team needs beyond that is being built now: shared boards, roles, handoffs. For that we are looking for a handful of companies that run makerOS in a real project.
Answered short and honestly. The rest in a conversation.
You can evaluate makerOS Solo for free, in commercial settings too. You only pay for your own AI subscriptions with each provider. makerOS Team will be a commercial product. We will define pricing and terms together with the pilot companies.
On your machines. makerOS needs no account, does not track, and contains no AI model of its own. Fully local development is possible, for example via Ollama. Your policy decides which models are connected.
Claude Code, Codex, Antigravity, Vibe (Mistral), Cline, OpenCode and Grok run as equal coding CLIs, including automatic MCP wiring. Plus every coding AI on the market, local models too, for example via Ollama. And you pick per role: build, verify, chat.
Through an engine, not through prompts. A definition of ready, fixed quality gates from plan to commit, and a Git hook that blocks the commit until the security agent reports green. Optionally, a second CLI reads along as an independent review partner.
macOS on Apple Silicon. The app is Apple-signed and notarized, an Intel build is in preparation. You bring your own CLI, installed and signed in.
The team edition is being built right now, together with the pilot companies. We deliberately name no date: what gets built first is decided by the pilots and their real needs.
macos (apple silicon) · your clis, your models · runs locally if you want