TS
Terms

AORKA · PUBLISHED REFERENCE · TOS-01

The agreements.

Plain language, structured clearly. These terms bind you as an individual user. Separate agreements bind the organization you work for, and are presented to an administrator who can sign for it. Read them. They matter.

—Terms of Service. Read it · Accepted by every person who uses the Service, on their own behalf.
—Trial License. Accepted for your organization when an evaluation begins. Superseded by the Customer Agreement on conversion. Read it in the console at app.aorka.com/terms.
—Customer Agreement. Accepted for your organization at signing by someone with authority to bind it. Read it in the console at app.aorka.com/terms.

First published March 29, 2026.

TOS
Document

AUDERAS INC · TOS-01 · REV. 2026.07

Terms of Service

Accepted by every person who uses the Service, on their own behalf.

Effective: July 30, 2026 · version 2026-07-30

01
Parties

Who you're agreeing with.

Aorka is operated by Auderas Inc, doing business as Aorka. Throughout these terms, “Aorka,” “we,” “us,” and “our” refer to Auderas Inc d/b/a Aorka. “Customer” refers to the organization using the Service. “User” refers to an individual person who accesses the Service under a Customer’s account. “You” and “your” refer to whichever of the two is accepting the document you are reading.

Three documents.

These terms are published as three documents drawn from a single source, each carrying its own version. Clauses common to more than one document are the same text in each.

—Terms of Service. The rules that apply to each person using the Service: eligibility, account security, acceptable use, and your responsibility for the commands you approve. Every user accepts these on their own behalf.
—Trial License. Governs an evaluation. A trial is a courtesy rather than a purchase, so this document sets no payment obligation, but it does carry the authorization warranty, liability limits, and indemnities that must be in force while real systems are connected. It is accepted for your organization when the evaluation begins.
—Customer Agreement. The commercial agreement: pricing and payment, intellectual property, service availability, and termination, alongside the same liability and indemnity terms. It is accepted for your organization at signing and supersedes the Trial License from that point.

Authority.

The Trial License and the Customer Agreement bind your organization, and accepting either is a representation that you have authority to bind it. The Terms of Service bind only the individual accepting them: doing so is not a representation of authority and commits your organization to nothing. A user without that authority is asked to accept the Terms of Service and nothing else.

02
Service

What the Service is.

Aorka is an IT infrastructure management platform. This includes:

—Remote endpoint management. A software agent runs on your endpoints with elevated privileges appropriate to the operating system. This agent executes commands dispatched through the platform. It connects outbound to Aorka's servers — no inbound ports are opened on your network.
—AI-assisted operations. An AI chat interface helps with troubleshooting, diagnostics, and script generation. AI outputs are not guaranteed to be correct. Every write operation passes through a three-layer safety pipeline (regex filter, AI risk evaluation, human approval) before execution. The human approval gate exists specifically because AI can make mistakes.
—Knowledge base and credential vault. A structured knowledge base stores facts about your infrastructure. A credential vault stores secrets with AES-256-GCM encryption. Both are scoped to your tenant and inaccessible to other customers.
—A public script library, with private parameters. Automation scripts are the deliberate exception to the isolation described above. Script bodies are held in a single library shared by every customer of the Service and are not partitioned by tenant. The values a script runs against — hostnames, account names, paths, and anything held in the vault — are supplied as parameters recorded against your own tenant-scoped records and do not form part of the shared body. The library is why the Service improves with use; the parameterization is what keeps your environment out of it. Do not defeat it by writing a value into the body of a script.

03
Accounts

Eligibility and accounts.

—Age requirement. You must be at least 18 years old to use Aorka. By creating an account, you represent that you meet this requirement.
—Account security. You are responsible for maintaining the security of your account credentials, MFA devices, and backup codes. Aorka authenticates through your Azure AD or Google identity provider — your organization's security policies (conditional access, MFA requirements, session controls) apply first. Aorka adds its own MFA layer for sensitive operations.
—Multi-tenant access. Aorka supports Microsoft Entra ID (Azure AD) and Google Workspace as identity providers. The first user to sign in from an organization not already known to us automatically becomes the tenant administrator for that organization. Subsequent users from the same organization require approval from that tenant administrator before gaining access. If your organization's Google or Microsoft directory allows accounts to be created without your approval, that first sign-in is the point at which control passes to whoever made it — you are responsible for managing user roles and access within your tenant from that moment.

04
Acceptable use

Acceptable use.

Aorka gives you powerful capabilities over real infrastructure. Use them responsibly.

You may not.

—Use Aorka to attack, compromise, or gain unauthorized access to systems you do not own or have authorization to manage.
—Attempt to circumvent, disable, or bypass the safety pipeline, script scoring system, or human approval gates.
—Share account credentials, MFA secrets, MCP tokens, or API keys with unauthorized parties.
—Use the platform to store, transmit, or process content that violates applicable law.
—Reverse-engineer, decompile, or attempt to extract the source code of the platform. The agent source code is plaintext and readable by design — this restriction applies to the server-side platform.
—Resell, sublicense, or provide access to the Service to third parties without written authorization.
—Intentionally inject prompts designed to manipulate the AI into generating harmful, deceptive, or unauthorized scripts.
—Place secrets, credentials, or confidential information in the body of a script. Scripts are held in a shared library that other customers of the Service can retrieve — use the credential vault and script parameters instead.

Authorization to manage connected systems.

The Service allows you to connect and manage endpoints, Microsoft 365 and other cloud tenants, network devices, and other systems (collectively, “Connected Systems”), and to execute actions that may read, modify, restart, or otherwise change them. Some Connected Systems may belong to your clients, customers, or other third parties. You represent and warrant that you have obtained all authorizations, consents, and legal rights necessary to connect those systems to the Service and to permit actions to be taken on them. The Service acts as a tool operating under your direction and approval; it does not independently decide what changes to make to your or your clients' systems.

Your responsibility.

You are responsible for all activity under your account. You are responsible for reviewing and approving scripts before execution. The human approval gate is a safety mechanism, not a rubber stamp. You are responsible for the commands you approve.

05
Safety pipeline

Safety pipeline and agent execution.

—The pipeline reduces risk; it does not eliminate it. Write operations pass through a three-layer safety pipeline: a deterministic pattern filter, an AI risk evaluation, and a human approval gate. Scripts scoring above 50 are blocked for all users with no override. Scripts scoring 50 or below may still cause unintended effects in your environment. The human approval step is the final and most important safeguard.
—Risk scores are not certifications. A risk score is an estimate produced by an AI model, and is subject to everything described in section 06. A low score means the automated evaluation found no obvious hazard in the text of the script. It is not a warranty that the script is safe, correct, or appropriate for the systems you are about to run it on.
—Agent execution. The Aorka agent executes commands with elevated privileges on your endpoints. While the agent is plaintext source code that you can inspect, and all commands pass through the safety pipeline, you accept the inherent risks of remote command execution on production systems.
—No guarantee of interception. We do not warrant that the safety pipeline will catch every dangerous command, every destructive side effect, or every command that is safe in general but harmful in your specific environment.

06
AI models

AI models and generated output.

Aorka's AI features are built on large language models supplied by third parties. This section describes the limits of what we can promise about them. It is important, and it is the section most likely to matter to you in practice.

—We do not build or control the models. We do not create, train, own, or control the AI models that power the Service. They are operated by third-party providers identified in the subprocessor list in section 07. Those providers may update, retrain, deprecate, rate-limit, or withdraw a model at any time, with or without notice to us. We may also change providers or model versions at any time. Model behavior can therefore change without any change to Aorka and without any action by you.
—We cannot attest that model output is reliable. We make no representation, warranty, or attestation — express or implied — that any model output is accurate, complete, current, safe, lawful, non-infringing, or fit for your environment or purpose. This applies to every kind of output the Service produces, including generated scripts and commands, diagnoses and root-cause explanations, risk scores, summaries, knowledge base entries, and recommendations. Models can and do produce output that is confidently stated and wrong. Identical inputs may produce different outputs on different runs.
—No human reviews output before you see it. Generated content is delivered to you automatically. It is not reviewed, verified, endorsed, or approved by Aorka personnel before it reaches you. Nothing the model produces constitutes professional, legal, security, or engineering advice from us, and no statement made by the model creates a warranty, commitment, or obligation on our part.
—You are responsible for reading and understanding what you run. Every script or command generated through the Service is presented to you in full, in plaintext, for review before it executes. By approving it you represent that you have read it, that you understand what it does and what it will change, and that you have determined it is appropriate for the systems it will run against. Do not approve a script you do not understand. Approval is the point at which responsibility for the execution and its consequences passes to you.
—Verify before you rely. You are responsible for independently verifying model output before relying on it, and for testing changes in a non-production environment where circumstances permit. You should not use the Service as the sole basis for any decision or action whose failure you could not tolerate.
—Your inputs are sent to the provider. Operating these features necessarily transmits your prompts, conversation history, connected-system data, and command output to the applicable model provider. Those providers are listed in section 07 and are bound by the terms described there.

To the maximum extent permitted by law, we disclaim all liability arising from model output, from your reliance on model output, and from any action executed on a Connected System that you or your users approved.

07
Subprocessors

Data processing and subprocessors.

Aorka uses third-party services to provide the platform. Your data may be processed by the following subprocessors in the course of normal operation:

—Anthropic. AI chat, script generation, and risk evaluation. Conversation data is sent via stateless API calls. Anthropic does not store your data or use it for model training.
—OpenAI. AI chat for accounts that select an OpenAI model. Conversation data is sent via stateless API calls. OpenAI does not use API data for model training by default.
—Voyage AI. Vector embeddings for semantic search across facts, scripts, and procedures. Text is sent via API to generate embeddings. Voyage AI does not store your content.
—Amazon Web Services. Infrastructure hosting (compute, database, secrets management). All data at rest is in AWS US regions. Database credentials are managed via AWS Secrets Manager.
—Stripe. Payment processing, subscription billing, and invoicing. Billing contact details and transaction records are transmitted to Stripe. Card details are entered directly with Stripe and are never received or stored by Aorka.
—SMTP2GO. Transactional email delivery for alerts and notifications. Only email addresses and notification content are transmitted.

We will provide 30 days' notice before adding a new subprocessor that processes customer data. Enterprise customers may negotiate additional data processing terms.

08
Export

Export controls.

Aorka is a US-based service. You acknowledge that the Service may be subject to US export control laws and regulations. You agree not to use, export, or re-export the Service in violation of any applicable export control laws, including US Export Administration Regulations (EAR) and Office of Foreign Assets Control (OFAC) sanctions. You represent that you are not located in, or a national or resident of, any country subject to US trade sanctions, and that you are not on any US government restricted party list.

09
Changes

Changes to these terms.

—Material changes. We will provide at least 30 days' notice before material changes to any of these documents take effect. Notice will be sent to the email address associated with your account, and the revised document will be published alongside the current version of every other document that governs your use of Aorka. Material changes include modifications to liability, data processing, pricing structure, or your rights.
—Changes we cannot wait 30 days to make. A change required to close a security risk, to comply with law or a regulator's direction, or to reflect a change imposed on us by a provider the Service depends on may take effect immediately or on shorter notice. We will still notify you, by the same means, and the notice will say why the period was shortened. We will not use this to change pricing, liability, or the limits on how we may use your data.
—Each document versions independently. Every document carries its own version identifier and changes on its own schedule. Revising the Customer Agreement does not require every user to re-accept the Terms of Service, and revising the Terms of Service does not require your organization to re-accept its agreement with us.
—Re-acceptance. We record which revision of each document was accepted, when, and by whom. When we publish a materially changed revision we may require the corresponding acceptance to be renewed before continued use — from each user for the Terms of Service, or from an authorized administrator for the Trial License or Customer Agreement.
—Non-material changes. Minor clarifications, formatting changes, or updates that do not affect your rights or obligations may be made without advance notice. The effective date shown at the head of each document will always reflect its latest revision.
—Acceptance. Your continued use of the Service after the effective date of a change constitutes acceptance of the updated terms. If you do not agree with a material change, you must stop using the Service before the change takes effect.

10
Disputes

Governing law and disputes.

—Governing law. These terms are governed by the laws of the State of Texas, without regard to conflict-of-law principles.
—Jurisdiction. Any disputes arising from these terms or the Service will be resolved in the state or federal courts located in Travis County, Texas. Both parties consent to the exclusive jurisdiction of those courts.

11
Contact

Contact.

Questions about these terms? Contact us at legal@aorka.com. Your use of the Service is also governed by our Privacy Policy.

Ready to get started?

See how Aorka manages infrastructure with AI safety built in.

Request a demo