Each answer has a stable link. Core is free and coming soon. Open source status has not been announced.
SayeOS Core is a free, governed workspace for persistent intelligent work. It is model, system, and connection agnostic, and public distribution is coming soon.
Yes. SayeOS Core is free.
SayeOS Core is free. Open source status has not been announced.
Not yet. Core is approaching public availability. Join the preview for release updates and access information.
Enter your email in the signup on the homepage. It opens a message to hello@sayeos.org.
Supported environment details are being finalized. Preview members receive environment updates first.
SayeOS is model agnostic and is designed to work with local models.
Yes. You can connect cloud model providers.
No. You connect and pay your model or token provider separately where applicable. SayeOS does not resell model access.
SayeOS coordinates work around your existing systems rather than requiring one application to become the source of everything. Systems of record keep their authority.
The work is not owned by one model provider. Different models can contribute, and changing providers does not force the work to restart.
SayeOS does not assume one universal way to reach systems, models, tools, or workers. Connections can include APIs, local tools, files, databases, approved services, and Edge mediated access.
Models, systems, and connections change. When the work does not depend on any one of them, it keeps its context and history through the change.
No. Your systems of record keep their authority. SayeOS coordinates work around them.
No. SayeOS references authoritative systems rather than absorbing their full datasets.
The workstream keeps its context, state, and history. Changing providers does not restart the work.
Yes. Different models can contribute according to task, cost, capability, privacy, environment, and policy.
A persistent unit of work that carries intent, context, state, evidence, decisions, and history across sessions and contributors.
A chat resets. A workstream continues, preserves its sources and decisions, and can be reviewed later.
Context is what matters for the work. State is what is currently true. Evidence is what supports a conclusion or action.
A bounded, executable capability with defined inputs, allowed tools, permitted actions, and review. It is not a saved prompt.
Models reason. Python executes defined work. Python is a first class execution layer, bounded by security, dependencies, and permissions.
Yes. People, models, Python workers, APIs, software services, and reviewers can contribute to one workstream.
Worker output is draft and pending review. A human review determines whether work is accepted.
So the rules governing consequential work are visible and versioned, and behavior can be checked against them.
No. Intelligence may analyze, simulate, and propose. It may not decide, approve, or substitute for human judgment.
A person can delegate scoped authority in advance. Systems execute inside that scope while it is valid and not revoked.
When an instruction is ambiguous or would require guessing intent, halting and escalating is safer than a confident wrong action.
Uncertainty is surfaced and escalated to a person, not resolved silently.
User data belongs to the user and is governed by standing protections. Governance records document decisions and actions and are retained for auditability.
Continuum preserves understanding over time without stripping away where it came from, what it meant, how well it is supported, or who has permission to share it.
No. It preserves understanding with source, context, confidence, and sharing boundaries, not a running log of everything said.
Individual, family and friends, entity, and public. They are sharing and publication boundaries.
An Understanding stays in its scope until it is moved outward through an explicit gate.
No.
Understanding shared within an organization or group, under that entity's own boundaries and review.
The Knowledge Engine works with documented knowledge. Continuum preserves human understanding. Facts are validated against sources or deterministic proof. Understandings preserve lived experience with source, context, confidence, disagreement, and permission attached.
Yes. A documented fact and a human Understanding may describe the same event from different perspectives. SayeOS can link them without silently treating them as the same thing.
No. Personal experience can be preserved as an attributed Understanding without being promoted into universal fact.
Yes. Evidence may support, contradict, narrow, or refine an Understanding. The history of that change stays visible.
Governed domain or organizational knowledge with known provenance and scope.
A bounded, executable capability packaged for reuse.
Context plus Skills plus configuration assembled for a class of problem.
A pack has been evaluated against defined criteria.
A higher, defined trust level, applied when applicable and not by default.
Yes. Builders, domain experts, and organizations can publish eligible offerings, and Saye publishes first party offerings.
No.