These Agent API Terms (“Agent Terms”) apply when you connect an AI agent, assistant or other software (“Agent”) to your Marginal AI account through the documented Marginal AI Agent API (“Agent API”). They supplement the general Terms and Conditions. Your contracting company is the Quantimental Technologies Ltd entity identified under the general Terms and in your applicable Order or purchase confirmation.
An applicable Enterprise Order may grant additional licence rights within its stated scope. The document hierarchy in the general Terms applies, including the priority of the Data Processing Agreement for covered personal-data processing and any mandatory transfer provisions. The Privacy Policy explains our personal-data processing.
1. Acceptance
You accept these Agent Terms when you approve your first Agent connection or accept them in Settings > Agent Access. We may update these Agent Terms. For a material change, we will explain the change and ask you to accept the new version before your Agents continue to use the Agent API, subject to applicable notice requirements and your existing contractual rights. A new version does not retrospectively remove accrued rights or change an agreed paid commitment without the contractual and legal protections that apply.
2. What the Agent API is
The Agent API allows an Agent you authorise to obtain Marginal AI intelligence and data on your behalf. Available tools include corporate and market events, figures reported in US Securities and Exchange Commission filings and macroeconomic series. Where available, research tools submit questions to the Marginal AI analysis service. The documented tools, inputs, availability and limits describe the scope of access.
Marginal AI supplies the information and research returned by these tools. It does not operate your separate Agent or execute that Agent’s downstream actions. Any change affecting agreed charges, included allowances or material contractual rights remains subject to the general Terms and applicable Order.
3. Your account, your Agents and your credentials
Connect an Agent only to an account you are entitled to use and within the authority you hold. An organisation’s access must remain within its subscribed seats and agreed API rights. You are responsible for the Agents you authorise and for protecting your credentials, without excluding our responsibility for our own breach or a liability that cannot lawfully be excluded.
Keep API keys confidential. Do not publish them, improperly share them or embed them in software accessible to persons who are not authorised to use them. If you believe a key or connection is exposed, revoke it using Settings > Agent Access and contact support as appropriate.
You can review and disconnect connected apps and revoke API keys in Settings > Agent Access. Logging out of all devices and closing your account also end your Agent connections, as described in the Service.
4. Permitted use
You may use the Agent API for lawful research, analysis and decision-making within the access and licence rights of your plan and any applicable Order. This includes automated retrieval and monitoring and use by an Agent you independently operate or choose.
You determine what to do with the information, including any lawful financial activity performed by you or your separate Agent. These Agent Terms do not require human review of every downstream action or prohibit otherwise lawful unattended activity by your independent systems. Marginal AI does not execute that activity or provide a brokerage, investment-advisory or discretionary-management service.
Your own professional activities, and any authorisations they require, remain your responsibility. This permission does not grant resale, redistribution, embedding or other licence rights beyond those expressly granted by your plan or Order.
5. Restrictions
You may not, or allow an Agent to:
- (a) use the Service unlawfully, including for market manipulation, insider dealing or an activity requiring an authorisation that you do not hold;
- (b) resell, redistribute, publish or systematically make underlying API data available to third parties, build or improve a competing dataset or service, or embed or white-label the Service, except to the extent expressly licensed in your Order and permitted by applicable third-party rights;
- (c) circumvent usage limits or seat restrictions, including by coordinated accounts, improper credential sharing or other avoidance measures;
- (d) access undocumented or unauthorised endpoints or probe, scan or test the security of our systems without our prior written permission; or
- (e) otherwise violate the applicable Acceptable Use Policy, data-protection requirements or third-party rights.
A customer’s lawful professional analysis using information it is entitled to access is not, by itself, a resale of the underlying feed. However, attribution alone does not create permission to redistribute protected material.
6. What the data is, and what it is not
The Agent API provides intelligence, data and research, including AI-generated analysis where available. Generated analysis may contain interpretations, estimates, errors or omissions and may differ on repeated requests. It is not a guarantee of accuracy, completeness, timeliness or an investment outcome.
Marginal AI does not provide investment advice or a personal recommendation assessed against your financial circumstances, investment objectives or suitability requirements. A response tailored to a requested topic, company, event, language or format is not offered as a suitability assessment. The financial-services disclaimer in the general Terms applies.
Marginal AI does not connect to brokerages, accept, place, route or execute orders, hold investment funds or exercise discretion over customer accounts. Any financial action is taken by you or your independent Agent with the relevant third party, not by Marginal AI.
Your Agents may misinterpret or act on information in ways you did not intend. You are responsible for choosing and configuring them and deciding what safeguards to use, subject to the laws and agreements applicable to your activities. This does not exclude our own contractual obligations or non-excludable liability.
7. Third-party AI providers
When you authorise an Agent to obtain information from Marginal AI, the requested information is made available to that Agent at your direction. Depending on the Agent you choose, it may be handled by you, your organisation or the Agent’s provider under the arrangements applicable to that Agent.
You are responsible for your relationship with an independently chosen Agent or provider and for compliance with its applicable terms. We do not guarantee its independent conduct. This does not remove our obligations for processing we carry out or for suppliers we appoint to provide our Service. Our processing is described in the Privacy Policy and any applicable Data Processing Agreement.
8. Sources and attribution
Returned information may include material from public filings and licensed third-party sources. Retain source attribution, links and notices where required when using or sharing material, and remain within the rights granted by your plan, Order and applicable third-party licences.
Do not misrepresent unreviewed AI-generated material as independently verified human-authored analysis. This does not prevent you from creating your own genuinely authored or reviewed research using information you are entitled to use. Attribution does not, by itself, grant redistribution rights.
9. Fees and limits
The Agent API is available on every plan. Each successful chargeable data-tool call consumes 0.5 CU on the same basis as other CU use, including any overage permitted by your plan. Lookup tools identified as free in the documentation remain free. A failed call is not charged. Calls are subject to the published per-minute and per-day limits.
Where the research tool is available, a research question consumes CUs on the same basis as a chat question of the same depth; a Deep Dive consumes more than a shallower question. The applicable balance and permitted overage terms apply. A request that cannot run within the applicable CU and usage conditions is not run.
The unpaid-overage limit in the general Terms or your Enterprise Order Terms applies to Agent usage as it does to other chargeable use. It limits unpaid charges, not total monthly expenditure. You remain responsible for applicable charges generated by Agents you authorise, subject to our own obligations and non-excludable rights.
A change to charging or included allowances will follow the specific notice and prepaid-commitment protections in the general Terms or Order, including at least the stated 30-day notice for changes to fees where applicable and any additional legal requirements. A general right to change documentation or tools does not override a fixed price or included allowance already agreed for a committed period.
10. Availability
The Agent API may be interrupted, rate-limited or changed, subject to our contractual obligations and applicable law. No additional uptime or service-level guarantee is created by these Agent Terms. Any expressly agreed service level or support commitment remains applicable. Descriptions such as “as is” and “as available” do not remove mandatory consumer rights or an express commitment in your agreement.
11. Suspension and revocation
We may suspend or revoke an affected key, connection or Agent API access where reasonably necessary to address a suspected compromise, security or stability risk, material breach or legal requirement. Where reasonably possible and lawful, we will explain the reason and any steps available to resolve it. Suspension and revocation remain subject to the agreement’s termination, refund and data-protection provisions.
12. Liability
The applicable liability and indemnity provisions in the general Terms and any Order govern the Agent API. We do not guarantee the conduct of independently chosen Agents or investment outcomes. To the extent permitted by law, we are not responsible for losses caused by your independent decisions or those Agents’ independent actions, including trading losses. This does not exclude responsibility for our own breach or liability that applicable law does not permit us to exclude. Business-only indemnities and caps do not override consumer rights.
13. Contact
Questions about these Agent Terms, and privacy requests, may be sent to support@marginal-ai.com.
