The rules for using Autonomous Relay Agents: what you may build, what you may not, how billing works, and who is responsible for what. The acceptable-use and prohibited-use clauses say what you may and may not build; the one most likely to catch you out is the clause on provider terms, because you route on your own keys and each provider’s policy binds you directly.
The agreement
These Terms of Service are a contract between you — or the organisation you represent — and Tech Service 4U Inc., a corporation incorporated under the laws of the Province of Manitoba, Canada, with its registered office at 17-955 Summerside Avenue, Winnipeg, Manitoba R3T 4Y1, Canada (“we”, “us”). They govern your use of Autonomous Relay Agents at autonomousrelay.com, the dashboard, and the API.
By creating an account, adding a credential, or making a request to the API, you accept these terms. If you are agreeing on behalf of a company, you confirm you are authorised to bind it, and “you” means that company.
Our Privacy Policy explains what we do with personal information and forms part of this agreement.
What Autonomous Relay Agents is, and what it is not
Autonomous Relay Agents is a routing and metering layer in front of AI model providers. You send one OpenAI-compatible request; we apply the rules you have set, select a provider endpoint that satisfies them, dispatch the request on a credential you supplied, stream the response back, and record what it cost.
We do not run models, and we do not resell inference. Every model is served on your own provider credential, at the provider’s own price, and we add nothing to the cost of a token. This is enforced in our database rather than promised in a policy.
What follows from that is the part people miss:
- Your contractual relationship for inference is with the provider, not with us.
- We cannot grant you rights to a model, raise your rate limits, or restore a provider account we do not control.
- If a provider suspends your account, the models you reached through it stop working here too, and we cannot appeal that on your behalf.
Your account
- You must be at least 18 and able to enter a binding contract.
- You must give accurate registration information and keep it current.
- You are responsible for everything done through your account, including by your team members and by software you authorise.
- One person or organisation per account. Do not share credentials to avoid paying for seats or to evade a limit.
You may not use the service if you are barred from doing so under Canadian sanctions law, or if you are located in a country subject to comprehensive Canadian sanctions.
Acceptable use — what you may do
The service is built for developers and businesses to run AI workloads. In plain terms, you may:
- Build and operate applications, agents and internal tools on top of the API.
- Use it commercially, including in products you charge for.
- Route requests across as many providers and models as your credentials allow.
- Automate your own usage, within the rate limits of your plan and of each provider.
- Export your usage data, ledger and invoices, at any time, in full.
- Evaluate and benchmark the service for your own purchasing decisions, and publish what you find.
We do not claim ownership of what you build, and we do not require attribution.
Prohibited use — what you may not do
This list is specific on purpose. A term banning “objectionable content” tells you nothing about whether your product is allowed.
Illegal content and serious harm
- Anything illegal under Canadian federal or provincial law, or the law where you or your users are.
- Child sexual abuse material, in any form. We report this to the National Child Exploitation Crime Centre and to Cybertip.ca, terminate the account immediately, and preserve records for law enforcement. There is no warning and no appeal.
- Non-consensual intimate imagery, or sexual content depicting real people who have not consented.
- Content that promotes, plans or provides material support for terrorism or mass violence.
- Instructions for building weapons, explosives, or chemical, biological, radiological or nuclear agents.
- Content that incites hatred against a group identifiable on grounds protected by the Canadian Human Rights Act, or that constitutes criminal harassment or uttering threats.
- Facilitating self-harm, suicide, or disordered eating.
Attacks on systems and people
- Malware, ransomware, exploit development for unauthorised use, or tooling whose purpose is to compromise systems you do not own.
- Phishing, credential harvesting, or impersonating a person, company or public body.
- Fraud, including synthetic identity fraud and generating fake reviews, documents or credentials.
- Spam, or any commercial electronic message that breaches Canada’s anti-spam legislation (CASL).
- Attacking our infrastructure or a provider’s: denial of service, unauthorised scanning or penetration testing, or deliberately circumventing rate limits, spend ceilings or guardrails.
- Reverse engineering the service, or accessing it to build a competing product from what you learn about its internals.
Privacy and surveillance
- Submitting personal information you have no lawful basis to process, or that you obtained by scraping in breach of a site’s terms.
- Biometric identification or face recognition of people who have not consented.
- Covert surveillance, stalkerware, or monitoring people without the disclosure their local law requires.
- Inferring sensitive characteristics — health, sexual orientation, religion, immigration status, union membership — about identifiable people without a lawful basis.
Consequential decisions about people
You may build these, but not on model output alone. Where output materially affects someone’s rights, safety or livelihood — employment, credit, housing, insurance, education, immigration, benefits, or anything medical, legal or safety-critical — you must keep a qualified human in the loop, disclose the automated processing where the law requires it, and provide a route to review. Quebec’s Law 25 and provincial human rights legislation apply to you here, not to us.
Deception at scale
- Political disinformation, astroturfing, or coordinated inauthentic behaviour.
- Presenting AI output as human-authored where a person would reasonably need to know, or as professional medical, legal or financial advice.
- Academic dishonesty services.
- Deepfakes of real people intended to deceive.
Commercial restrictions
- Reselling access to the API as a general-purpose gateway, or repackaging it as a competing routing service. Building a product that uses the API is expected and encouraged; reselling the routing layer itself is not.
- Using outputs to train a competing model, where the provider whose model produced them forbids it.
- Sharing an API key outside your organisation.
Reporting abuse. If you believe someone is using Autonomous Relay Agents in breach of this clause, write to admin@tecs4u.com.
Provider terms apply to you directly
This is the clause most likely to catch you out, so it gets its own heading.
Because you route on your own credentials, you are the provider’s customer. OpenAI’s, Anthropic’s, Google’s and every other provider’s usage policies bind you directly, whether or not you have read them. Their restrictions are frequently stricter than ours, and they change without telling us.
- A use permitted by these terms may still breach a provider’s policy, and their policy wins for traffic sent to them.
- Providers may suspend or terminate your account with them for their own reasons. We have no control over that and no ability to reverse it.
- Where a provider requires you to disclose AI involvement, label output, or restrict a use case, that obligation is yours.
We publish each endpoint’s declared data policy in the catalog so you can check before routing, but we do not warrant that a provider’s published policy is accurate or current. Verify anything you intend to rely on.
Your keys and your security
- Keep your Autonomous Relay Agents API keys secret. Anyone holding one can spend against your account.
- Rotate a key immediately if it may have been exposed. You can revoke keys yourself at any time.
- You are responsible for charges incurred through your keys before you revoke them, including by someone who obtained one without your permission.
- Set spend ceilings. They exist precisely so a runaway agent or a leaked key has a bounded cost, and we consider using them ordinary care.
We protect provider credentials you store with us as described in the Privacy Policy, clause 4. We never return a stored credential to anyone, including you.
Plans, credits, billing and taxes
All prices are in US dollars.
| Plan | Price | What you pay for |
|---|---|---|
| Solo | Free. Credit top-ups carry a purchase fee, disclosed at checkout. | The routing layer. Inference is on your own provider key at their price. |
| Team | US$99 per month, 0% on BYOK usage | Scoped keys, spend ceilings, PII redaction, health-aware routing. |
| Enterprise | Custom, invoiced annually | SSO/SAML and SCIM, private routing tier, extended ledger retention, contractual SLA. |
Credits
- Credits are prepaid and are spent against usage. They are not a deposit and not a currency, and they carry no cash value outside the service.
- Credits do not expire while your account is open and in good standing.
- We do not extend credit unless it is written into an Enterprise agreement. Without one — which is every account by default — a request is refused when it would take your balance below zero. Where an agreement sets a negative limit, that figure is shown as credit_line on your credits endpoint and the same refusal applies at that point instead.
- A balance can still end up negative, and you owe the difference. We reserve an estimate before a request runs and settle the real cost afterwards, and a provider can report more usage than we reserved — a stream already in flight cannot be cut off part-way. When that happens we debit the true amount rather than absorb it. While a balance is negative every request is refused, including free models, until you top up.
- Every debit is reconstructable from the ledger, which you can export at any time.
Refunds
Credits are non-refundable. Once purchased they can be spent on the service, but not returned as cash. Subscription fees are charged in advance and are not refunded for a partial month.
There are two exceptions, because both are ours to cause rather than yours:
- If we discontinue the service, or materially reduce it under clause 10, we refund your unused credits.
- If we change these terms to your material disadvantage and you close your account rather than accept them, we refund your unused credits.
Nothing in this clause removes a right you have under provincial consumer protection legislation that we cannot contract out of — including The Consumer Protection Act (Manitoba). Where such a right applies, it applies despite anything above.
Taxes
Prices exclude tax. We charge GST/HST, QST or provincial sales tax where required, based on your billing address. You are responsible for any other taxes arising from your use. Business customers outside Canada may be required to self-assess.
Your content and intellectual property
You keep everything you send and everything you get back. We claim no ownership of your prompts, your outputs, or what you build. We do not use them to train anything, and we do not store them at all unless you switch on logging within your own organisation.
You grant us only the licence needed to run the service: to transmit your request to the provider you selected and return the response to you. That licence ends when the request does.
We keep ownership of Autonomous Relay Agents itself — the software, the catalog, the routing logic and the brand. Nothing here transfers any of it to you.
About model output
Ownership of AI output is unsettled law in Canada and elsewhere, and it varies by provider and by jurisdiction. We make no representation that output is original, that it can be protected by copyright, or that using it will not infringe someone else’s rights. Get your own advice before relying on output commercially.
Availability and changes
We work to keep the service available and fast, and we publish provider health so you can see when a provider is not. Free and Team plans carry no contractual uptime commitment; Enterprise agreements may include an SLA, which then governs.
Much of what affects your experience is a provider’s uptime, not ours. Our failover routing is designed to reduce that exposure, not eliminate it.
We may change, add or remove features. If we materially reduce a feature you rely on, or discontinue the service, we will give you at least 30 days’ notice and let you export your data before access ends.
Suspension and termination
You may close your account at any time. On closure we stop billing you, and you can export your data first. Unused credits are not refunded on a voluntary closure — see clause 8.
We may suspend or terminate your account if:
- You breach these terms, particularly the prohibited uses in clause 5.
- Your use threatens the security, integrity or availability of the service or a provider.
- Payment fails and is not resolved after we ask.
- We are required to by law.
Where the circumstances allow, we give notice and an opportunity to fix the problem first. For serious harm — child sexual abuse material, an active attack, a legal order — we act immediately and explain afterwards.
Credits are non-refundable on termination, as clause 8 sets out — except that if we terminate your account without cause, or discontinue the service, we refund your unused credits. We do not refund where we terminated for a breach of clause 5. Clauses on liability, indemnity and governing law survive.
Disclaimers
Except as expressly stated here, the service is provided “as is”. To the extent the law permits, we disclaim implied warranties of merchantability, fitness for a particular purpose and non-infringement.
We specifically do not warrant that:
- Model output is accurate, complete, current, lawful or fit for any purpose. AI systems produce confident, plausible and wrong answers, and you must verify anything that matters.
- Any provider will remain available, keep its prices, or continue to offer a model.
- A provider’s published data or retention policy is accurate.
- The service will be uninterrupted or error-free.
Nothing in this clause limits rights you have under provincial consumer protection legislation. Where that legislation gives you a warranty or remedy that cannot be excluded, it applies despite anything above.
Limitation of liability
To the extent the law permits, neither party is liable for indirect, incidental, special or consequential damages, or for lost profits, revenue, data or goodwill, even if told such damages were possible.
Our total liability arising out of or relating to this agreement is limited to the greater of the amounts you paid us in the 12 months before the claim, or US$100.
These limits do not apply to our gross negligence, wilful misconduct or fraud, to death or personal injury caused by our negligence, or to any liability that cannot be limited under Canadian or applicable provincial law — including consumer protection legislation.
Charges you incur with a model provider are between you and that provider. We are not liable for them, including where a bug, an agent loop or a compromised key caused them — which is why spend ceilings exist and why we treat using them as ordinary care.
Indemnity
You will indemnify us against claims, losses and reasonable legal costs arising from your use of the service in breach of these terms, from content you submit, from your breach of a provider’s terms, or from your infringement of another person’s rights. We will tell you promptly of any such claim and will not settle it without your consent.
Governing law and disputes
These terms are governed by the laws of the Province of Manitoba and the federal laws of Canada that apply there, without regard to conflict-of-laws rules — Manitoba being both where we are incorporated and where we operate. The courts of Manitoba have exclusive jurisdiction, and each party submits to them.
If you are a consumer, nothing here deprives you of the protection of mandatory provisions of the law of your own province or country of residence, or of your right to bring proceedings there where the law gives you that right. Manitoba consumers retain their rights under The Consumer Protection Act(Manitoba), and Quebec consumers the recourse available under the Consumer Protection Act (Quebec).
Before starting proceedings, please contact us and give us 30 days to resolve the matter. Most disputes end there.
Language
These terms are published in English and in French. Under the Charter of the French Language as amended by Bill 96, a contract of adhesion — which these terms are — must be presented to a Quebec customer in French first; the customer may then expressly choose another language. Use the language switcher in the header or footer of any page.
[The French version is a careful translation but has NOT been certified. Have it reviewed by a qualified legal translator before relying on it in Quebec — the French version is the one a Quebec customer is entitled to rely on]
If you find a discrepancy between the English and French versions, please report it to admin@tecs4u.com so it can be corrected.
Changes and general terms
We may update these terms. We will post the new version with a new date, and for material changes we will give notice by email or in the dashboard at least 30 days before they take effect. Continuing to use the service after that means you accept them; if you do not, close your account and we will refund your unused credits under clause 8.
- If a provision is held unenforceable, the rest stands and that provision is narrowed to the minimum extent needed.
- Not enforcing a term once does not waive it.
- You may not assign this agreement without our consent; we may assign it in a merger or sale of the business.
- These terms and the Privacy Policy are the entire agreement between us about the service, and replace anything said earlier.
- Neither party is liable for failures caused by events outside its reasonable control.
Contact
Tech Service 4U Inc.
17-955 Summerside Avenue, Winnipeg, Manitoba R3T 4Y1, Canada
General enquiries: admin@tecs4u.com
Abuse reports: admin@tecs4u.com
Privacy: see the Privacy Policy, clause 16.