Stellary Beta Is Open
The Stellary beta is live: delivery, documents, cockpit dashboards, AI agents, MCP, automations and API in one workspace. Early teams can apply via Discord.
6 min read
Stellary was built around a simple idea: teams should not have to split execution, context, and AI across five different tools.
Project work is no longer just cards on a board. Real teams need a shared place for delivery, documentation, decisions, priorities, automations, and agents that can act inside the actual system instead of floating above it.
This introduction was reviewed against the current product on July 27, 2026.
Why We Built Stellary
Most teams still operate with a fragmented stack.
- tasks in one tool
- docs in another
- decisions in chat
- status in meetings
- AI in a separate assistant with shallow context
That fragmentation slows teams down in ways that are easy to underestimate. People lose time reconstructing context, cross-checking status, and translating the same information between systems.
We wanted a product where the work, the context around the work, and the AI layer all live in the same operating environment.
What Actually Lives Inside Stellary
Stellary is structured around a real operating model:
- organizations
- workspaces
- projects
Inside that structure, teams can work across several connected layers.
Delivery
The board handles the execution layer:
- cards
- statuses
- scopes
- priorities
- delivery views
This is where work gets tracked and moved forward day to day.
Documentation
The knowledge layer keeps documents close to execution rather than off in a separate wiki.
Specs, references, runbooks, ADR-style material, and working context are easier to keep alive when they sit next to the project instead of outside it.
Piloting
The cockpit adds the layer that classic ticket tools usually miss:
- missions
- priorities
- decisions
- proposed actions
That is the difference between simply tracking tasks and actively piloting a project.
AI Runtime
Stellary also includes a real AI surface:
- setup assistance
- agents
- MCP access
- plugins
- skills
- pipelines
- automations
The goal is not to bolt AI onto the side. The goal is to let AI work with the real project context, under the right permissions and approvals.
What Makes Stellary Different
Three things matter most.
1. Context is first-class
AI is only useful when it can work with the real state of the project.
That means:
- actual cards
- actual documents
- actual missions and priorities
- actual workspace rules and permissions
Without that, most AI workflows collapse back into summarization.
2. Delivery and piloting belong together
Traditional tools are good either at tracking execution or at documenting strategy. Teams still have to glue the rest together themselves.
Stellary is built so the tactical layer and the strategic layer inform each other continuously.
3. MCP is part of the product, not an afterthought
Through MCP, compatible clients and agents can connect to Stellary as a real working system.
That matters because AI should not just read about the project. It should be able to understand and participate in the workflow in a governed way.
How Teams Use It
The most natural use cases tend to look like this:
- run delivery inside a project with cards, scopes, and documents
- keep decisions and priorities visible in the cockpit
- connect agents through MCP for research, execution support, or structured proposals
- extend the workspace with plugins, skills, pipelines, and automations
- use the API reference when a team wants programmatic control
This is especially useful for teams that have started to outgrow the split between PM tooling, docs tooling, and AI tooling.
Getting Started
The fastest way to understand Stellary is to follow the docs in this order:
If you want the broader product update, read Stellary Beta Is Open.
What Comes Next
We are still early, but the direction is clear.
Stellary is not trying to become "another tool with an AI button." We are building an operating system for teams that want delivery, context, and agents to work together inside the same environment.
