# Terms of Service **Effective Date:** April 28, 2026 **Last Updated:** April 28, 2026 --- ## 1. Introduction Welcome to Mitrai. These Terms of Service ("Terms") govern your access to and use of all Mitrai products and services, including the **Mitrai Memory Engine** (self-hostable RAG engine) and **Mitrai Cloud** (hosted AI chat application). By accessing or using any Mitrai service, you agree to be bound by these Terms. If you do not agree, do not use our services. "Mitrai," "we," "us," and "our" refer to **Mitrai, operated by Lakshmi Arumugam**, contactable at **legal@mitrai.live** ("Mitrai"). "You" and "your" refer to the individual or entity accessing our services. "Service" or "Services" refers collectively to all Mitrai products, websites, APIs, and related offerings. --- ## 2. Products and Services ### 2.1 Mitrai Memory Engine The Mitrai Memory Engine (version 0.2.0, MIT License) is a self-hostable, production-ready RAG (Retrieval-Augmented Generation) memory engine that you install and run on your own infrastructure. It provides three interfaces: - **REST API** — 15 endpoints on port 8100 for integration with applications and agent frameworks - **MCP Server** (`memory-mcp`) — compatible with Claude Desktop, Cursor, and any MCP-compatible AI agent - **CLI** (`memory-engine`) — setup, health checks, document ingest, and query from your terminal **Distribution:** Install via `pip install mitrai-memory` (PyPI) or via Docker at `ghcr.io/mitrai/memory-engine`. **AI Provider:** The Memory Engine routes all AI operations through LiteLLM and supports multiple providers including NVIDIA NIM, OpenAI, Anthropic, Cohere, HuggingFace, and Ollama. **The default AI provider is NVIDIA NIM**, using the `nvidia_nim/nv-embed-v1` embedding model and `nvidia/nemotron-3-nano-30b-a3b` optimizer LLM. When the default NVIDIA NIM configuration is active, document chunks and queries are transmitted to NVIDIA's API. A fully local, zero-external-call deployment is achievable by configuring Ollama as the provider. **Storage:** Uses Qdrant for vector storage (per-user collections prefixed `mem_`) and local filesystem for training data. **Your Responsibility:** When self-hosting the Memory Engine, you are responsible for hosting, securing, configuring, and maintaining your deployment, including any data stored within it. ### 2.2 Mitrai Cloud Mitrai Cloud is a hosted AI chat application built on the Chainlit framework (Python/React), available at [mitrai.web.app](https://mitrai.web.app). Mitrai operates the backend infrastructure on your behalf. Features include: - AI chat with multiple models and modes (general chat, reasoning, vision/multimodal) - Built-in document upload and retrieval-augmented generation - Optional Google Drive integration - Persistent chat history **AI Provider:** Mitrai Cloud uses **NVIDIA NIM as its primary and default AI provider** via the API at `integrate.api.nvidia.com/v1`. Models deployed include: - General chat: `qwen/qwen3.5-122b-a10b`, `minimaxai/minimax-m2.7` - Reasoning: `minimaxai/minimax-m2.7` - Vision/multimodal: `google/paligemma` (via NVIDIA VLM endpoint) - Embeddings: `nvidia/nv-embed-v1` **IMPORTANT — Data Processing:** When you use Mitrai Cloud, your prompts, messages, and uploaded document content are transmitted to NVIDIA's API servers for processing. This is the default behavior. See Section 6 (Third-Party AI Providers) for full details. **Infrastructure:** User credentials, chat history, and license data are stored in PostgreSQL. Document vector embeddings are stored in Qdrant. Uploaded document files are stored on Mitrai's servers. --- ## 3. Free Trial New Mitrai Cloud accounts receive a **14-day free trial** that activates automatically upon account registration. No credit card is required to start the trial. During the trial period, you have full access to all Mitrai Cloud features. At the end of the 14-day trial period, your account automatically converts to the free tier. You will not be charged unless you affirmatively choose and activate a paid plan. --- ## 4. Subscription and Billing ### 4.1 Free Tier After your trial expires, your account converts to the free tier, which includes limited usage of Mitrai Cloud features. Specific limits are described on our pricing page. ### 4.2 Paid Plans Paid plans provide expanded usage limits and additional features. Pricing is published at [mitrai.web.app](https://mitrai.web.app). By subscribing to a paid plan, you authorize Mitrai to charge your designated payment method on a recurring basis according to the plan cycle (monthly or annual, as selected). ### 4.3 Cancellation You may cancel your paid subscription at any time through your account settings. Cancellation takes effect at the end of the current billing period. No partial refunds are issued for unused time in a billing period unless required by applicable law. ### 4.4 Downgrades If your usage at time of downgrade exceeds the limits of the lower tier, access to excess data or features may be restricted. We will attempt to notify you in advance before restricting access. ### 4.5 Price Changes We reserve the right to change pricing. We will provide at least 30 days' notice before any price change affects existing paid subscribers. --- ## 5. Permitted and Prohibited Use ### 5.1 Permitted Use You may use our Services for lawful purposes consistent with these Terms, applicable law, and our Acceptable Use Policy. ### 5.2 Prohibited Use You may not use our Services to: - Violate any applicable law, regulation, or third-party rights - Generate, distribute, or store content that is illegal, abusive, defamatory, obscene, or constitutes hate speech - Attempt to reverse-engineer, decompile, or extract source code from the Mitrai Cloud application - Use automated means to scrape, crawl, or systematically extract data from our services without written permission - Circumvent, disable, or interfere with security features, rate limits, or access controls - Use our Services to train competing AI models or to build substantially similar competing services - Impersonate any person or entity, or misrepresent your affiliation with any person or entity - Transmit malware, viruses, or any other malicious or destructive code - Engage in activity that imposes an unreasonable or disproportionately large load on our infrastructure - Violate the terms of any third-party AI provider whose services we use (including NVIDIA's usage policies) ### 5.3 Content Rights and Responsibility You may not upload, connect, or process any content through our Services unless you hold all rights, permissions, and consents necessary for Mitrai to store that content, generate embeddings from it, and transmit it to NVIDIA or any other configured AI provider for processing. By submitting content, you represent and warrant that you have all required rights and consents for this chain of processing. ### 5.4 Restricted Data Unless Mitrai has expressly agreed in a separate written agreement, you must not use the Services to process: - Protected Health Information (PHI) or any data subject to HIPAA or equivalent health privacy laws - Payment card data subject to PCI DSS - Biometric data - Special-category personal data under GDPR (Article 9), including racial or ethnic origin, political opinions, religious beliefs, health data, or genetic data - Export-controlled or ITAR-restricted information - Information subject to attorney-client privilege, professional confidentiality, or court order restrictions The default configuration transmits your content to NVIDIA's API; such regulated data categories are not appropriate for this processing path. --- ## 6. Third-Party AI Providers ### 6.1 NVIDIA NIM (Default Provider) **Both Mitrai Cloud and the default configuration of the Mitrai Memory Engine use NVIDIA NIM as the primary AI provider.** This is not an optional feature — it is the default and standard operating mode. When you use these services under the default configuration: - Your prompts, messages, queries, and the content of any documents you upload or process are **transmitted to NVIDIA's API servers** at `integrate.api.nvidia.com/v1` for processing. - NVIDIA processes this content to generate AI responses and embeddings. - NVIDIA's own privacy policy, terms of service, and data processing agreements govern NVIDIA's handling of your data. You can review NVIDIA's privacy policy at: [https://www.nvidia.com/en-us/about-nvidia/privacy-policy/](https://www.nvidia.com/en-us/about-nvidia/privacy-policy/) By using Mitrai Cloud or the Memory Engine in its default NVIDIA NIM configuration, you expressly instruct and authorize Mitrai to transmit your content to NVIDIA for processing on your behalf. If you deploy the Memory Engine for use by other users (e.g., your employees or customers), you are responsible for providing those users with equivalent notice and obtaining any consents required by applicable law before their content is processed by NVIDIA or any other third-party AI provider you configure. ### 6.2 Fully Local Option (Ollama) If you require that no content leave your infrastructure, you must explicitly configure the Memory Engine (self-hosted) to use **Ollama** as the AI provider. When configured with Ollama and a local Qdrant instance, 100% of processing occurs on your own machine and no data is transmitted externally. Mitrai Cloud does not currently offer a fully local configuration. ### 6.3 Other Providers Self-hosted Memory Engine deployments may be configured to use other third-party AI providers including OpenAI, Anthropic, Cohere, or HuggingFace. When using any of these providers, your content is transmitted to that provider's API. Each provider's own terms and privacy policy govern their handling of your data. ### 6.4 No Liability for Third-Party Providers Mitrai is not responsible for the data handling, availability, or conduct of any third-party AI provider. We strongly encourage you to review the applicable privacy policies and terms of service of any provider whose API processes your data. --- ## 7. Google Drive Integration Mitrai Cloud offers an optional Google Drive integration. This feature is: - **Strictly optional** — you must explicitly authorize it - **OAuth-based** — you authenticate via Google's OAuth flow; Mitrai stores an OAuth access token in your user account metadata to enable recurring access - **Scoped to your authorization** — Mitrai only accesses files you explicitly select or authorize When Google Drive is connected, selected files are downloaded to Mitrai's servers and treated as uploaded documents for RAG processing (including transmission to NVIDIA's API for embedding). Your use of the Google Drive integration is also subject to Google's Terms of Service and Privacy Policy. You may revoke Mitrai's access to your Google Drive at any time through your Google account settings or through Mitrai's account settings page. --- ## 8. API Usage ### 8.1 Rate Limits Mitrai may impose rate limits on API calls, document uploads, and other operations. Specific limits depend on your plan tier. Exceeding rate limits may result in temporary throttling or suspension of your access. ### 8.2 Upload Limits (Memory Engine) The Memory Engine enforces the following default upload limits (configurable by the self-hosting operator): 50 MB per file, 20 files per batch, 200 MB per batch total. ### 8.3 API Key Security You are responsible for keeping any API keys or credentials issued by Mitrai confidential. Do not share keys publicly or embed them in client-side code. If you believe a key has been compromised, rotate it immediately via your account settings. Mitrai is not liable for unauthorized use of credentials due to your failure to keep them secure. ### 8.4 Suspension Mitrai reserves the right to throttle or suspend API access without notice if we detect abuse, excessive usage, or violations of these Terms. --- ## 9. Data Processing ### 9.1 Memory Engine (Self-Hosted) When you self-host the Memory Engine, you control your own infrastructure and data storage. - **Default NVIDIA NIM configuration:** Document chunks and queries are sent to NVIDIA's API. NVIDIA processes this content subject to their terms. - **Ollama configuration:** All processing is fully local. No data leaves your machine. - **Other cloud providers (OpenAI, Anthropic, etc.):** Content is transmitted to the configured provider. Mitrai does not receive or store your Memory Engine data in the self-hosted configuration. You bear full responsibility for the security and compliance of your self-hosted deployment. ### 9.2 Mitrai Cloud Mitrai Cloud is a hosted service. When you use Mitrai Cloud: - Your account information (email, hashed password, metadata) is stored on Mitrai's servers in PostgreSQL. - Your chat history is stored in PostgreSQL on Mitrai's infrastructure. - Documents you upload are stored on Mitrai's servers. - Document vector embeddings are stored in Qdrant on Mitrai's infrastructure. - Your prompts, messages, and document content are transmitted to NVIDIA's API (the default AI provider) for processing. We apply industry-standard security practices to protect data stored on our infrastructure, including encryption in transit (TLS) and access controls. See our Privacy Policy for full details. --- ## 10. Intellectual Property ### 10.1 Your Content You retain ownership of all content you submit, upload, or create through our Services ("User Content"). By submitting User Content, you grant Mitrai a limited, non-exclusive, royalty-free license to store, process, and transmit your User Content as necessary to provide the Services. ### 10.2 Mitrai Memory Engine The Mitrai Memory Engine software is open-source and licensed under the **MIT License**. Your rights to use, modify, and distribute the Memory Engine are governed by the MIT License. ### 10.3 Mitrai Cloud The Mitrai Cloud application, its source code, design, branding, and all associated intellectual property are proprietary to Mitrai. You are granted a limited, non-transferable, non-exclusive license to access and use Mitrai Cloud as a service. You may not copy, distribute, or create derivative works from the Mitrai Cloud application. ### 10.4 Mitrai Trademarks "Mitrai" and associated logos are trademarks of Mitrai. You may not use our trademarks without prior written permission. --- ## 11. AI Output Limitations AI-generated content is produced by probabilistic language models and may be inaccurate, incomplete, outdated, or otherwise unreliable. Mitrai makes no representations or warranties regarding the accuracy, reliability, or fitness for purpose of any AI-generated output. You are solely responsible for reviewing, verifying, and relying on any AI output. Do not use AI output as the sole basis for medical, legal, financial, or other critical decisions. --- ## 12. Disclaimers THE SERVICES ARE PROVIDED "AS IS" AND "AS AVAILABLE," WITHOUT WARRANTY OF ANY KIND. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, MITRAI EXPRESSLY DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, AND ANY WARRANTIES ARISING FROM COURSE OF DEALING OR USAGE OF TRADE. We do not warrant that the Services will be uninterrupted, error-free, secure, or free of viruses or other harmful components. We do not warrant that defects will be corrected. --- ## 13. Limitation of Liability TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, IN NO EVENT WILL MITRAI, ITS AFFILIATES, OFFICERS, DIRECTORS, EMPLOYEES, AGENTS, OR LICENSORS BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, PUNITIVE, OR EXEMPLARY DAMAGES, INCLUDING LOSS OF PROFITS, DATA, GOODWILL, OR OTHER INTANGIBLE LOSSES, ARISING OUT OF OR IN CONNECTION WITH YOUR USE OF OR INABILITY TO USE THE SERVICES, EVEN IF MITRAI HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN NO EVENT WILL MITRAI'S TOTAL AGGREGATE LIABILITY TO YOU FOR ALL CLAIMS ARISING FROM OR RELATED TO YOUR USE OF THE SERVICES EXCEED THE GREATER OF (A) THE AMOUNTS YOU HAVE PAID TO MITRAI IN THE TWELVE (12) MONTHS PRECEDING THE CLAIM OR (B) ONE HUNDRED DOLLARS (US $100). SOME JURISDICTIONS DO NOT ALLOW THE EXCLUSION OR LIMITATION OF CERTAIN DAMAGES. THE ABOVE LIMITATIONS MAY NOT APPLY TO YOU TO THE EXTENT PROHIBITED BY APPLICABLE LAW. --- ## 14. Modifications to Terms We reserve the right to modify these Terms at any time. When we make material changes, we will: - Update the "Last Updated" date at the top of this document - Provide notice via email (to the address associated with your account), in-app notification, or prominent notice on our website, at least 14 days before changes take effect for material modifications Your continued use of the Services after the effective date of updated Terms constitutes your acceptance of the revised Terms. If you do not agree to the revised Terms, you must stop using the Services. --- ## 15. Account Termination ### 15.1 Termination by You You may delete your Mitrai Cloud account at any time through your account settings. Upon deletion, we will cease providing Services to you and will delete or anonymize your data in accordance with our Privacy Policy and data retention schedule. ### 15.2 Termination by Mitrai We may suspend or terminate your access to the Services at any time, with or without notice, if we determine that: - You have violated these Terms or our Acceptable Use Policy - Your account presents a security risk or legal liability - Continued provision of Services to you is no longer commercially viable ### 15.3 Effect of Termination Upon termination, your right to use the Services ceases immediately. Provisions of these Terms that by their nature should survive termination will survive, including Sections 10 (Intellectual Property), 11 (AI Output Limitations), 12 (Disclaimers), 13 (Limitation of Liability), 14 (Indemnification), and 17 (Governing Law). --- ## 16. Indemnification You agree to defend, indemnify, and hold harmless Mitrai and its affiliates, officers, directors, employees, and agents from and against any claims, liabilities, damages, losses, and expenses (including reasonable legal fees) arising out of or relating to: - Your User Content, including any claim that it infringes third-party intellectual property or privacy rights - Your violation of these Terms or any applicable law or regulation - Your failure to obtain necessary consents from third parties whose data you process through our Services - Your use of our Services in a way that causes harm to any third party - Your operation of a self-hosted Memory Engine deployment, including any privacy or data protection claims arising from your deployment This indemnification obligation survives termination of your account or these Terms. --- ## 17. Governing Law These Terms are governed by and construed in accordance with the laws of the **Province of Ontario** and the federal laws of **Canada** applicable therein, without regard to conflict of law provisions. Any dispute arising from or relating to these Terms or the Services shall be subject to the exclusive jurisdiction of the courts of Ontario, Canada. If you are accessing the Services from outside Canada, you agree that your use is subject to Canadian law and the courts of Ontario, and you waive any objection to such jurisdiction. --- ## 18. Contact Information For legal inquiries: **legal@mitrai.live** For support: **support@mitrai.live** For privacy concerns: **privacy@mitrai.live** Mitrai | mitrai.web.app --- *These Terms of Service are effective as of April 28, 2026.*