Frequently asked questions, answered concisely.

Agents can retrieve company knowledge with sources, read decisions and their status, submit knowledge and decision drafts, and request policy checks on planned actions. The interface also supports exception requests, proposed decision relationships and status checks. Accessible capabilities depend on the client, permissions and installation.

No. NomOS MCP capabilities do not require a NomGate service. NomOS checks access to its own content and functions. An action assessment through MCP alone does not stop external shell commands or model calls. NomGate is planned to add local control for Windows and macOS.

Agents submit knowledge, decisions and relationships as proposals. Responsible people review and approve them in NomOS. An agent’s proposal does not automatically become a binding basis.

NomOS makes company knowledge, current policies and reviewed decisions available to employees and connected AI agents. Your team can use AI for specific tasks and build on reviewed work in similar cases.

Reviewed results and decisions, with their sources, rationale and scope. Tokens already spent are not recovered. Reuse can avoid repeated research and model work; retrieval, review and processing still consume resources.

Technical checks and blocking apply to execution paths integrated for that purpose. We test their coverage in the pilot. Your organisation remains responsible for policies and business decisions.

With SaaS, Ainomos operates NomOS in the agreed cloud environment. With Self-Hosted, your organisation runs it in your own infrastructure. In both cases, processing by connected models depends on the model configuration and agreed requirements.

Model requests on paths connected to NomOS pass through its gateway. The gateway controls which approved context may be sent to the configured model provider. Both SaaS and Self-Hosted can connect compatible external or locally hosted models. The actual processing location depends on the selected model endpoint.

NomOS organises the impact graph into separate rooms. Each room has its own policies, approved content and an accountable owner. A question is answered only with knowledge approved for that room. Departments, mandates or clients remain separated, and every access is assigned to the correct identity and room context.

Yes. Security, access, policy and agent events can be streamed to the customer’s SIEM. The implementation defines which events, fields, formats and retention periods apply. The corresponding evidence also remains traceable in NomOS.

Yes. NomOS supports single sign-on through OpenID Connect (OIDC) in both SaaS and Self-Hosted. The customer’s existing identity provider remains the source for identities and authentication.

NomOS does not use company content to train a generally available language model. Whether a connected model provider uses inputs for its own purposes depends on the selected service and contract. Model classes, routing and the operating model determine which data may reach which endpoints.

No. LLMs can be replaced without rebuilding the impact graph. Knowledge, policies, decisions, corrections and evidence remain independent of the model currently in use. A new model can therefore continue from the same reviewed foundation. A new model connection requires compatibility and quality checks. Changing the search model may also require rebuilding the search index.

Ainomos is the company. NomOS is our first product. NomGate is planned as a local control tool for Windows and macOS. NomApp is intended to provide mobile knowledge access, notifications and approvals. NomStore is announced as a knowledge exchange platform. These extensions are not yet generally available.

From October 2026, NomOS enters production use with selected customers. Ainomos supports the rollout for both SaaS and Self-Hosted. General availability (GA) follows in March 2027.

NomOS includes web chat, MCP connectivity, human approvals and knowledge maintenance. The pilot tests the functions and integrations you need. These may include SSO, SIEM, Dreaming and technical action controls. The agreed scope determines the configuration.

No. Dreaming creates proposals for maintaining knowledge and policies. People review and approve changes. Responsibility always remains with people.

Dreaming requires a connected large language model (LLM). For curating the impact graph, we recommend locally hosted models in a controlled environment. This can be combined with SaaS or Self-Hosted. Suitability, interfaces, hardware and licensing are assessed for your process.

NomOS offers a web chat and gives employees and agents access to shared company knowledge. It connects sources, policies and earlier decisions. On connected paths, it checks access and agent actions and records the context used and approvals given.

Employees use NomOS Chat on the web. NomApp is planned as a mobile extension. AI agents connect through the open MCP protocol. Both sides work with the same impact graph, policies and traceable approvals.

Both SaaS and Self-Hosted can connect LLMs through supported interfaces. Customers place models into configurable classes, for example by sensitivity, data location, cost or capability. Each task defines which model classes are eligible. The gateway routes a request only to a model in an approved class.

NomOS imports Markdown files, for example from Confluence or Obsidian, as well as PDF and Word files. File import does not continuously synchronise with the source system. We assess further integrations and required write permissions against your systems during the pilot.

NomOS can assess a submitted action against identity, room, policies and required approvals. The result can include permission, a caution, an approval requirement or a denial. Technically blocking an external action requires its execution path to enforce that decision.

Sources, policies and decisions carry their validity, provenance and accountable owner. Superseded content and rejected proposals remain marked as such instead of returning as open options. Conflicts are exposed for review rather than silently merged.

An evidence record identifies the basis used for a case, how an action was assessed and who approved it. Event coverage, signing, export and retention must match the specific installation. We establish these technical requirements before operation. A citation alone does not establish that an answer is factually correct.

No. NomOS integrates with your existing identity and access infrastructure, and your identity provider remains the source for identities. NomOS adds rooms, policies and approvals for knowledge, models, tools and agent actions.

Identify the job your current tools do not handle well enough: do several teams need to establish the same policies, approve exceptions or reuse earlier decisions? NomOS provides a shared basis for authorised teams in those workflows. If your current system already covers this well, you do not need another platform for it.

Subject-matter owners approve content, review changes and resolve conflicts. IT maintains identity and integrations. NomOS supports this work; it does not replace subject-matter responsibility. The pilot measures ongoing maintenance effort alongside potential benefit.

NomOS offers its own workspace and interfaces for connected agents and business applications. The pilot defines the access channels and integrations actually needed for two workflows. This does not promise automatic compatibility with any application.

The pilot takes 30 days and covers discovery, two live workflows and a joint evaluation. We agree the schedule for a subsequent rollout based on the sources, responsibilities and integrations needed. Those results can then be reused for similar requests.

You select two concrete, recurring workflows and name an accountable business owner. We also need access to the relevant sources and a technical contact for identity and integrations. A company-wide migration is not required for the pilot.

Price and scope depend on the operating model and requirements. The pilot has a fixed price and agreed scope.

That depends on your workflows. NomOS can reduce time spent searching, explaining and correcting, increase throughput and preserve work already completed. Counting time saved and additional throughput in full would count the same benefit twice. The calculator under “Value and effort” uses your assumption about the reusable share. The pilot compares two real workflows with your current approach, including setup and ongoing maintenance.

NomOS is the central AI workspace for employees. In the web chat, they receive the relevant context, sources and applicable policies. They review proposals and correct them when needed. People retain responsibility and approval, while decisions, rejections and corrections return to the impact graph with a traceable record.

Business Support is available Monday to Friday, 09:00 to 18:00 Swiss time, excluding public holidays in Zug. We provide support in German and English by email and through a ticketing system. Callbacks and video appointments are arranged as needed.

For a critical production outage with no practical workaround, we respond within two support hours. When a key feature is substantially affected, we respond within four support hours; for other bugs and usage questions, within one support business day. Response times are measured during support hours. A person takes ownership, assesses the impact and identifies the next step. This is an initial response, not a guaranteed resolution time. For critical outages, we provide updates every two support hours until service is stabilised.

During the first 30 days, you have a named contact and a weekly meeting. Together, we review the rollout, open questions and next steps. The 30-day pilot covers two real workflows at a fixed price with an agreed scope.

For SaaS, Ainomos operates NomOS and handles monitoring, updates and backups. Scope and restoration arrangements are agreed with you. For Self-Hosted, your organisation operates the infrastructure and backups; Ainomos assists with NomOS, configuration, troubleshooting and updates. Further operational assistance, custom integrations, data preparation and additional training are agreed separately.

For business-critical use, an additional on-call service can be agreed. It covers critical production outages around the clock, with a qualified initial response within one hour. This additional service is not automatically included in Business Support.