# Legion Intelligence > Secure, agentic AI for mission-critical systems > **Legion Intelligence** is the full name of the platform, and throughout this document, we’ll refer to it simply as **Legion**. Legion delivers secure, AI agents that integrate directly into high-stakes workflows across sectors like defense, aerospace, and manufacturing. Teams can interact with Legion through a single, secure interface—or embed assistants directly into the tools they use every day, like ERP systems, mission dashboards, or supply chain platforms. This flexibility ensures fast adoption with minimal disruption. Unlike generic copilots, Legion connects the dots across fragmented systems and data stores, enabling users to search, chat, and create with the full power of their organizational knowledge—no matter where it lives. ## Built for security-minded teams, Legion supports IL2–IL6 deployments, air-gapped networks, and total control over data, models, and infrastructure. ## Core Capabilities ### Search **Find What Matters—Fast** Search your entire organization’s knowledge base—across systems, formats, and silos—using natural language or keywords. Over 100 secure integrations and powerful proprietary [Retrieval-Augmented Generation (RAG)](https://www.legionintel.com/blog/rag-enterprise-ai-advancements) system to deliver instant, context-aware results. ### Chat **Ask Questions, Get Real Answers** Use RAG to query your internal documents and systems in everyday language. Users can access consolidated, source-linked intelligence to support faster, better decision-making. ### Create **Turn Knowledge into Content** Generate emails, reports, proposals, or briefs using Agentic AI Workflows. Quickly create and standardize high-quality content, from routine emails to complex reports, with built-in citation support. ### Automate **Scale Smarter With AI Agents** Deploy no-code, AI-powered agents to handle repetitive workflows—such as summarizing documents, pulling data, and automating ticketing systems—freeing your team to focus on higher-value tasks. > \*Agentic AI Workflows: Task-specific, multi-step AI sequences that operate autonomously—yet securely—with human-in-the-loop awareness, integrating into your systems to gather data, execute processes, and produce outcomes." --- ## Who Legion is For We are built for security-minded teams tackling the world’s most complex challenges: from government agencies, both Defense and Federal, handling national security to enterprise leaders in aerospace, manufacturing, engineering, and financial services. We exist to serve members of the workforce and teams who need AI that enhances—not disrupts—their mission. ## These teams operate in high-stakes environments, where precision, speed, and security aren’t optional—they’re mission-critical. ## Legion Value Proposition Legion is a secure, enterprise-ready AI orchestration platform built for sensitive, high-stakes environments. It enables organizations to unlock actionable intelligence, accelerate complex workflows, and orchestrate autonomous agents—all without relinquishing control over data or systems. Designed for teams that cannot compromise on security or outcomes, Legion puts trust, speed, integration, and operational control at the core. Whether supporting mission planning, MRO, procurement, or intelligence analysis, Legion delivers precise AI assistance where and when it's needed most—inside the systems teams already use. Organizations choose Legion to move faster, stay secure, and achieve real operational results with AI that works the way they do. --- ## For Technical Buyers Legion is a secure, enterprise-scale generative AI platform that enables you to unlock the power of your organizational data rapidly without compromising security, auditability, or technical requirements. Trusted by the US Department of Defense (DOD) and Department of Energy (DOE), our secure modular architecture functions like a pre-assembled system where components can be swapped as your needs evolve—ensuring your investment remains future-proof as AI advances. **Key Pillars:** - **Secure by Default** - Data never leaves your environment; LLMs do not train on it—not even during inference. - SOC 2, HIPAA, and GDPR compliant - Deployable from IL-2 through IL-6 (Controlled → Classified SECRET) - **Flexible by Design** - Use any LLM: Open-source or proprietary - Connect to 100+ platforms (CRM, storage, communication tools, etc.) - Deploy anywhere: On-Prem, Cloud, VPC, SaaS, Edge—even Bare Metal - Supports Model Context Protocol (MCP) for agent-based workflow automation - **Control with Clarity** - Track and prioritize compute, token use, and LLM selection - Role-based and user-level access control for data, agents, and models - Minimal infrastructure requirements (e.g., runs efficiently on just 2 GPUs) - **Rapid Results** - Deploy in weeks, not quarters - Monthly platform updates - Maintainable with as little as one part-time engineer --- ## Government & Defense Users Legion helps military and intelligence teams modernize how they process information, plan missions, and execute operations—without compromising on security. Whether deployed in classified defense networks or across civilian federal systems, Legion delivers AI-native workflows and assistants that meet the highest operational standards. **Mission-Aligned Capabilities:** - **Seamless Access to Information** - Search and summarize across dozens of file types using natural language - Deployable in NIPR and SIPR (Secure/Classified) environments (IL-2 → IL-6 compliant) - Supports both structured (e.g., databases, spreadsheets) and unstructured (e.g., reports, emails) classified document environments. - **Improved Decision-Making** - All answers are traceable to verified sources - Simplify complex topics from acronyms to advanced intel - Instantly find documents by title, keyword, or embedded content - **Operational Efficiency** - Streamline briefs, SITREPs, and reports using prebuilt templates - Merge multiple intelligence sources into a single, digestible view - Ask questions or follow-ups conversationally—no training required ## Federal Mission Teams Federal civilian agencies—ranging from the DOJ,DOE, DOS to the VA—are facing increasing demands for speed, accountability, and insight, all while managing legacy systems and sensitive data. Legion equips these mission teams with AI agents that work within existing tools and architectures to reduce overhead, increase visibility, and unlock data-driven decisions. **How Legion Supports Federal Missions:** - **Rapid Answers from Complex Systems** - Federate knowledge across case systems, compliance databases, emails, and more - Search across agencies or divisions with role-based permissions and audit trails - Improve public-facing response times without risking internal data leakage - **Automation Where It Matters** - Generate reports, findings, FOIA responses, and regulatory summaries instantly - Configure prebuilt workflows with no-code logic and approval gates - Cut administrative burden while maintaining full human-in-the-loop control - **Secure by Design** - SOC 1-2 type 2, HIPAA, GDPR, and CMMC-aligned controls - Flexible deployments: cloud, hybrid, on-prem, or classified - Configurable model governance to meet compliance and procurement requirements Legion gives federal teams the tools to act quickly, confidently, and securely—without ripping and replacing core systems. The result: faster delivery on mission goals, better service to the public, and smarter decisions powered by data. **Industry-Specific Solutions:** - [Federal](https://www.legionintel.com/government/federal-industry): Empower federal agencies to unlock insights, streamline operations, and increase oversight without data exposure. - [Defense](https://www.legionintel.com/government/defense-industry): Support mission-readiness with secure, agentic AI tailored to the needs of military operations and national defense. **Featured Use Cases:** - [Mission Planning](https://www.legionintel.com/government/use-cases/mission-planning): Rapidly synthesize logistics, terrain, and intel data to create dynamic, responsive mission plans. - [Secure Process Automation](https://www.legionintel.com/government/use-cases/secure-process-automation): Reduce manual overhead in regulated workflows with AI agents that operate in secure enclaves. - [Intelligence Analysis](https://www.legionintel.com/government/use-cases/intelligence-analysis): Consolidate unstructured data from multiple sources and produce decision-grade intelligence with citations. --- ## End User Enterprise Legion helps enterprise teams unlock the full value of their organizational knowledge—without sacrificing control or security. By connecting fragmented data sources, automating routine workflows, and embedding AI assistants and agents directly into the tools teams already use, Legion transforms how work gets done in high-stakes industries. **Enterprise-Grade Capabilities:** - **Seamless Access to Information** - Unified search across all data systems and formats - Trusted by national security agencies for sensitive deployments - Simple natural language interface requires minimal onboarding - **Improved Decision-Making** - Transparent, verifiable responses grounded in internal data - Clarify technical and complex topics quickly - Avoid the hassle of data migration by utilizing federated search across your existing systems. - **Time Savings** - Automate emails, reports, and internal documentation - Summarize and extract insights from large data sets - Ask follow-up questions from a single, unified place, without switching tools or screens **Industry-Specific Solutions:** - [Aerospace](https://www.legionintel.com/enterprise/aerospace-industry): Streamline compliance, accelerate R&D, and ensure traceability across engineering and operations. - [Manufacturing](https://www.legionintel.com/enterprise/manufacturing-industry): Enhance efficiency with AI-driven search, quality control, and supply chain optimization. - [Engineering](https://www.legionintel.com/enterprise/engineering-industry): Quickly access technical documentation, blueprints, and regulations to reduce cycle times. - [Financial Services](https://www.legionintel.com/enterprise/financial-industry): Accelerate due diligence, automate compliance, and support complex reporting across regulatory environments. **Featured Use Cases:** - [Intelligent Procurement](https://www.legionintel.com/enterprise/use-cases/intelligent-procurement): Centralize purchasing intelligence, reduce cost leakage, and surface strategic insights from procurement data. - [Modern MRO](https://www.legionintel.com/enterprise/use-cases/modern-mro): Speed up maintenance and repairs by surfacing manuals, procedures, and inventory data through natural language queries. Legion ensures your enterprise doesn’t just add AI—it integrates it where it matters most: at the point of work, inside the tools and processes your teams rely on daily. --- ## How We're Different - [Security-First Design](https://www.legionintel.com/security): Built for secure/confidential zero-trust environments; protect sensitive data at every level - LLM Agnostic: Choose and route across models that meet your operational, security, or performance needs. Legion enables **granular orchestration—even at the task level within a single workflow—**so teams can leverage the best model for each specific job, without vendor lock-in. - [Modular Architecture](https://www.legionintel.com/blog/the-journey-to-secure-generative-ai-balancing-fort-knox-level-security-with-flexibility): Integrate seamlessly with what you already use—no need to rip and replace --- ## Learn More - [Platform](https://www.legionintel.com/platform): Dive deeper into how Legion’s secure, agentic AI platform works across architectures, environments, and workflows. - [Blog](https://www.legionintel.com/blog): Explore case studies, deployment insights, and customer success stories showcasing transformative AI solutions across industries. - [Careers](https://www.legionintel.com/careers): Join Legion to help shape the future of mission-critical AI and contribute to securing vital processes across the globe. - [Company](https://www.legionintel.com/company): Learn more about Legion’s expertise, values, and approach to delivering secure, agentic AI solutions. - [Contact](https://www.legionintel.com/contact): Talk to our team and explore how Legion can fit your specific environment and unique use cases. --- ## Frequently Asked Questions (FAQs) **Q: What is Legion?** A: Legion is an agent-orchestration platform for defense, government, and enterprise organizations. It helps teams work faster and more efficiently by connecting to all their mission-critical apps, systems of record, and trusted data sources. It enables users to search, chat, write, and take action on their organization’s knowledge through AI-powered tools. The Legion platform is highly flexible, allowing organizations to customize key capabilities such as the underlying AI models and directly integrate with Legion’s powerful retrieval-augmented generation (RAG) engine. **Q: What is Legion's Vision?** A: Legion’s vision is to accelerate human impact. **Q: What is Legion's Mission?** A: Legion’s mission is to create a world where disciplined AI is at the heart of every operation, secures the nation, and powers universal progress. **Q: Is Legion a chatbot?** A: No, Legion is not just a chatbot. It is a platform of AI-driven tools—including enterprise search, chat, content creation, and workflow automation—that extend far beyond basic conversational capabilities. With Legion, users can find answers, draft emails, generate reports, and significantly improve operational efficiency. **Q: What is the difference between Yurts and Legion?** A: Our rebrand from Yurts to Legion reflects our growth and deepened focus on mission-critical environments. Like a Roman legion, our platform unifies intelligent AI agents to empower teams across defense, government, and enterprise. This evolution marks our transformation from a flexible AI platform to an active force of AI assistants, delivering actionable intelligence and transformative results. We remain committed to secure integration, operational adaptability, and meeting your evolving needs. This change represents a renewed dedication to solving your most pressing challenges—with a legion of intelligent agents to help you save time and work more effectively. **Q: I built a chat app using ChatGPT for my company. Why would I need Legion?** A: GPTs are great for personal projects, but they aren’t built to scale for large or complex use cases. Organizations require enterprise-grade features to successfully deploy Generative AI at scale. That’s where Legion comes in. Legion offers over 100 out-of-the-box integrations with your company’s applications—including the ability to build custom integrations. It allows you to embed assistants into the systems your teams use every day and provides IT and Data Governance teams with the tools they need to ensure enterprise control and airtight security—trusted even by the U.S. Department of Defense. Legion is also cost-efficient, helping to reduce potentially massive compute expenses. **Q: What are the capabilities of Legion?** A: Legion has a wide range of capabilities. A few include: - **Search**: Retrieve information from diverse sources - **Chat**: Manage documents, policies, and manuals - **Write**: Draft emails, reports, and proposals - **Summarize**: Summarize meeting notes, emails, and research - **Generate**: Create FAQs, reports, and descriptions - **AI Assistants**: Integrate Legion directly into your applications - **Monitor**: Track performance with real-time metrics and dashboards - **Security and Privacy**: Role-based access controls, SSO, encryption, and mission-critical security controls - **Private and Secure Deployment**: Deploy into any environment including VPCs, on-prem, air-gapped networks, and IL-2, IL-4/5, and IL-6 environments **Q: What is the difference between Legion and SOFchat?** A: Legion is the company that built SOFchat. SOFchat is powered by Legion intelligence. Legion is a secure AI platform that enables coordinated agentic workflows across complex, mission-critical systems. It helps teams work faster and more efficiently by connecting to all your mission-critical apps, systems of record, and trusted data sources. It enables you to search, chat, write, and take action on your organization’s knowledge through AI-powered tools. The Legion platform is incredibly flexible, allowing organizations to customize key capabilities such as the underlying AI models and directly integrate with Legion's powerful retrieval augmented generation (RAG) engine. **Q: How does Legion help special operations teams?** A: Legion powers SOFchat, enabling special operations forces to rapidly access and act on critical information. It improves decision-making, increases time savings, and delivers intelligence securely—without compromising compliance or confidentiality. **Q: Is Legion secure enough for sensitive military information?** A: Yes. Legion is secure by default and trusted by the U.S. Departments of Defense and Energy. It maintains data sovereignty by acting as a secure intermediary between your data and the LLM of your choice—ensuring your information never leaves your environment and is never used to train external models. Core security features include: 1. Secure deployment options including on-premises, Virtual Private Clouds (VPCs), and air-gapped systems—approved for IL-2, IL-4/5, and IL-6 environments 2. Role-Based Access Control (RBAC) 3. Single Sign-On (SSO) and Multi-Factor Authentication (MFA) **Q: Is Legion approved for Secret/Confidential Information?** A: Yes. Legion is approved for use in classified environments, including IL-2, IL-4/5, and IL-6 deployments. **Q: Where can Legion be hosted or deployed?** A: Legion can be deployed anywhere—including on-prem, cloud, hybrid, VPC, SaaS, edge, and even bare-metal environments—maintaining complete data sovereignty while meeting IL-2 through IL-6 compliance for government-grade security. **Q: How does Legion handle information access?** A: Legion provides instant, verifiable answers from your documents using natural language search, allowing you to access all your data in one place—when it matters most. **Q: How does Legion provide verifiable answers?** A: Legion searches for and prioritizes answers in your organizational data and includes source citations with every response, enabling confident decision-making based on trusted information. **Q: Can Legion help with administrative tasks?** A: Yes. Legion streamlines admin work using AI to draft emails, create SITREPs, and handle routine paperwork—freeing up time to focus on high-priority missions. **Q: What makes Legion different from other AI platforms?** A: Legion is purpose-built for mission-critical operations. Its modular architecture integrates seamlessly into existing systems and secure environments—without requiring risky migrations. **Q: What industries or teams use Legion?** A: Legion is used by both enterprise and federal agencies. Key industries include: 1. Aerospace 2. Manufacturing 3. Supply Chain Management 4. Engineering Teams 5. Financial Services 6. Government (Defense and Civilian sectors) **Q: Who is using Legion currently?** A: Legion is trusted by the U.S. Department of Defense (DoD) and the Department of Energy (DoE). **Q: How does Legion improve decision-making?** A: Legion cuts through noise to surface relevant, verified intelligence—whether you're writing reports, planning missions, or gathering operational insights—so you can make smarter decisions, faster. **Q: Is Legion customizable to our specific needs?** A: Yes. Legion is highly flexible—you can customize integrations, select your preferred LLMs, and adapt features to your workflows without sacrificing security or performance. **Q: What kind of administrative controls does Legion offer?** A: Legion provides deep administrative visibility including: - Resource allocation and compute monitoring - Prioritization controls - Granular access controls across documents, agents, and LLM usage **Q: How does Legion handle data retrieval?** A: Legion connects to your mission-critical applications and systems of record, using its Retrieval-Augmented Generation (RAG) engine to return context-rich results from across disparate data sources. **Q: What types of workflows can Legion enhance?** A: Legion transforms mission-critical workflows across defense, government, and enterprise—from research and planning to admin, intelligence, and automated reporting. **Q: Is Legion future-proof as AI technology evolves?** A: Yes. Legion's modular architecture is built to evolve with AI—allowing you to swap components as your operational and technological needs change. **Q: How does Legion protect sensitive information?** A: Legion maintains full data sovereignty. Your information never leaves your environment and is never used to train third-party AI models. **Q: Does Legion work with multiple AI (LLM) models?** A: Yes. Legion is LLM-agnostic, letting you use the models that best fit your compliance, security, and operational needs—without vendor lock-in. **Q: How would I typically use Legion in my day-to-day?** A: For general users—especially in government or special operations—Legion supports you by: 1. **Seamless Information Access:** Access all your data in one place. Legion provides instant, verifiable answers from your documents using natural language search. 2. **Administrative Task Streamlining:** Use AI to draft emails, create SITREPs, and handle paperwork, freeing time for mission-critical work. 3. **Decision Support:** Legion cuts through noise to surface critical intelligence needed for informed decision-making, whether planning operations or writing reports. **Q: How can I try Legion?** A: Contact us and we'll have you up and running in no time. **Q: What is Legion's strategy around Agentic AI (Agents)?** A: Legion AI assistants proactively deliver actionable intelligence in mission-critical environments. They enable coordinated AI workflows across systems while preserving security and data sovereignty. Agents operate within your infrastructure, keeping users in control and enabling informed, timely decisions. **Q: How does Legion keep my data secure / ensure data security?** A: Legion uses a multi-layered security strategy: - **Data in transit** is encrypted with TLS. - **Data at rest** is protected with AES-256 encryption in access-controlled repositories. - A **secure middle-man architecture** ensures data sovereignty—your data stays in your environment and is never used to train AI models. Legion supports secure deployments including: - On-premises - Hybrid environments - Private cloud - Air-gapped networks - IL-2 through IL-6 environments (up to classified SECRET) Legion is also SOC 2 compliant and trusted by the DoD and DoE. **Q: What connectors / data integrations does Legion offer?** A: Legion integrates with over 100 enterprise systems and data sources. If your system has an API—even a legacy or custom system—Legion can likely connect to it. **Q: I don’t see my application listed as an integration with Legion. What data can you integrate with?** A: Legion supports custom integrations with any application. Contact us to discuss connecting your specific system or data source. **Q: What are the key product features of Legion?** A: - **Chat:** Ask questions or generate content using internal or web data. Responses are secure, and no data is used to train LLMs. - **Search:** Use natural language to find information across your organization's data. Get verified answers fast. - **Agents / Assistants:** Automate workflows to reduce manual work and improve productivity. - **Shortcuts:** Helps users quickly get started and discover use cases across the platform. **Q: What are Legion's advantages over NIPRGPT?** A: In addition to multi-model support (use any LLM of your choice), Legion/SOFchat offers: 1. Support for IL-6 (Secret / Classified) data—usable on both NIPR and SIPR networks 2. Advanced search across document content, statements, metadata, titles, and tags 3. 100+ available integrations to help automate workflows 4. Embedded AI assistants deployable on any web page or internal site 5. More than double the supported data types—including audio formats **Q: What are Legion's advantages over CAMOGPT?** A: Alongside multi-model support, Legion/SOFchat offers: 1. More robust search capabilities across content, metadata, and structure 2. 100+ enterprise integrations for workflow automation 3. Embedded assistants usable across web properties 4. Support for 3x more data types—including audio formats **Q: How does Legion compare to ChatGPT?** A: ChatGPT is not designed for secure or sensitive environments. Legion/SOFchat lets you safely use AI with secure documents and classified data while maintaining control over your environment. Your data never leaves your infrastructure, is never used for training, and works with any LLM you choose. Results are verifiable, accurate, and fast. **Q: How does Legion compare to AskSage?** A: In addition to multi-model serving, Legion/SOFchat offers: 1. Support for IL-6 (Secret / Classified) data on both NIPR and SIPR 2. Deep search capabilities across document content and metadata 3. 100+ integrations to streamline your workflows 4. Embedded assistants usable on any internal or external web platform **Q: What logs does Legion have access to? How does it handle logs from internal users?** A: Legion stores performance and system logs **only in your environment**. These logs are accessible to you and used by the Legion team solely to improve onboarding and user enablement. **User logs are never used to train or fine-tune LLMs.** **Q: How does Legion work?** A: Legion connects directly to your existing systems and data sources, turning AI into a seamless, secure layer of your workflow. It works with top LLMs—like GPT-4, Claude, Mistral, and Llama—enhanced by Legion’s retrieval augmented generation (RAG) engine. You can deploy it flexibly: in the cloud, on private servers, or at the edge. Key features include: - Role-based access control - In-platform analytics - Model hub for customizing and switching LLMs **Q: If I use Legion, will everyone have access to all my information?** A: No. Legion supports **role-b** **Q: What happens if a user wants to delete a document or collection?** A: Collections can be created by a system administrator or assigned to a user by an admin or another user. The creator of the collection has the ability to add to, delete, or curate the collection. System administrators always have the ability to edit or modify collections. Collections may also be private to an individual user, in which case that user has access to manage and curate their collection as they see fit. **Q: What information do you need from me to set up a demo instance in the cloud?** A: To set up a cloud-based demo (sandbox), we need: - Number of users: How many users will access the demo? - Timeline: Duration of the demo period - Use Case / Objective: Specific goals for the demo - Hosting: Where you plan to host (our cloud vs. yours) - Integrations: Any systems or applications needing integration We’ll also schedule a scoping call to clarify the goals of your Proof of Concept (POC). **Q: Are there any fees for setting up a demo instance?** A: We typically provide a 30-day trial period. If you choose **Legion-hosted cloud**, the demo is free of charge. If you require an **on-premise or self-hosted** demo, a $20,000 fee applies to cover engineering labor costs. **Q: Can you provide an example of the agreement we need to sign upon payment of fees?** A: Yes. We will provide a sample agreement that outlines: - Scope of Work - Payment terms - Project duration - Terms and responsibilities for both parties This ensures alignment before project kickoff. **Q: Are there any restrictions on the demo? Can my colleagues run their own prompts?** A: We recommend no more than 10 users to ensure optimal support. Yes, your colleagues can run their own prompts. We encourage collaborating with our Customer Success team for the best experience and adoption. **Q: How long does it take to deploy a Legion POC?** A: Deployment time depends on your hosting preference: - **Legion Cloud:** ~48 hours - **External / On-premise:** ~1 week **Q: What is the current pricing for Legion? How much does Legion cost?** A: Pricing depends on a few factors: - **Platform Fee:** Based on environment complexity - **Per-User Fee:** ~$45 PUPM (pricing varies) - **Customization Fee:** Based on your unique needs - **Hosting / Compute Fee:** Depends on your preferred deployment environment **Included in your purchase:** - Dedicated Customer Success Manager (CSM) - User enablement and adoption training - Full technical support **Q: How do I know where the information comes from when I get a response?** A: Our platform provides citations for the sources of information used in responses. These citations can be based on your data or internet sources if enabled. Legion's sources panel can be expanded to give you direct access to the documents from which the information was derived, allowing you to verify the accuracy and trustworthiness of the response. **Q: How do you safeguard ITAR-sensitive or other sensitive data from being accidentally sent out during Internet searches?** A: Our system includes guardrails that prevent users from asking sensitive questions. These guardrails apply to model chat, Retrieval-Augmented Generation (RAG), and internet searches. This ensures that sensitive data, such as ITAR-sensitive information, is protected from accidental exposure. **Q: What does Legion provide when a team wants to self-host the solution?** A: When a team opts to self-host the solution, we provide Terraform scripts to facilitate the setup. If the team already has a cluster, we can utilize that. If they need a cluster built from the ground up, we can assist with that as well. All components are containerized within a Kubernetes (KB) cluster. We have experience deploying to high-security environments, including government cloud. Our team is available to help with the deployment process, or if the team prefers to handle it themselves, that is also fine. **Q: What level of support do you provide for updates and bug fixes?** A: Legion's support services are available during standard business hours, defined as 6:00 AM to 5:00 PM Pacific Time, Monday through Friday, excluding U.S. federal holidays. The level of support, including response times, varies depending on the severity of the incident: - **Urgent (SEV1):** Response within 2 business hours - **Low (SEV4):** Response within 5 business days **Q: Do you have public Service Level Agreements (SLAs)?** A: Our Service Level Agreements (SLAs) are tailored to meet the specific needs of our customers. We scope the SLAs based on the requirements and expectations of each customer to ensure they receive the level of service that best supports their business objectives. **Q: What support does Legion offer?** A: Legion offers several types of support to ensure your success: - **User and Administrator Training:** Training sessions for both end users and administrators to ensure they are well-equipped to use the platform effectively. - **Documentation:** Detailed documentation to guide users through various features and functionalities. - **Custom Workflow Support:** Assistance in creating and managing custom workflows tailored to your specific needs. - **Engineering Support:** Support for deployment, API integrations, system administration, updates, and more to ensure smooth operation and integration with your existing systems. **Q: How does Legion ensure that prompts, referenced indexed data, and outputs do not leave my environment and are not used to train when using proprietary models like OpenAI's Chat GPT, Anthropic's Claude, etc.?** A: Legion offers flexibility in choosing which LLMs to deploy. If you use a proprietary model vendor like OpenAI or Anthropic, data specific to your request will be sent to the model’s provider to process your request, subject to the terms of your agreement with the provider. Note that most providers do not train on your data if you have an enterprise agreement with them. For the most secure deployments, you can use open-source LLMs hosted on-premises, allowing for air-gapped deployment. These are often preferred by security-conscious organizations to ensure data does not leave your environment. **Q: How does Legion's approach to Secure AI differ from Glean and their "Cloud Prem" model, where Glean is deployed as a managed service in isolation within your own Private Cloud (VPC) environment?** A: Legion, even when deployed with proprietary LLMs like Anthropic's Claude or OpenAI’s GPT-4, is significantly more secure than Glean's “Cloud Prem” model for a number of reasons: - **Glean’s “Cloud Prem” Model:** Still provides Glean with full access to your data for indexing and searching, significantly increasing the surface area of the data leaving your environment to be processed by their LLM. - **Legion’s Approach with Proprietary LLMs:** Legion processes, indexes, and searches your data within your environment, only sending the minimal required data to the proprietary cloud-based LLM of your choice (e.g., Anthropic's Claude or OpenAI’s GPT-4) to formulate the answer to your question. The LLM does not have access to search, index, or train on your data directly, minimizing the data that leaves your environment. - **Legion’s Approach with Open-Source LLMs:** For maximum security, you can use Legion with an open-source LLM (e.g., Mistral or Llama). This option ensures that your data is fully processed within your environment and your data will never leave your four walls. Our most security-conscious users choose this approach for their deployment for a truly air-gapped solution. **Q: What is Legion Model Hub?** A: Legion's Model Hub is a central repository to manage models, model runtime parameters, and register additional models (open source, fine-tuned by organizations, etc.). **Q: Can you tell me about Autoscaling?** A: Autoscaling is an automated service that dynamically adjusts the number of GPU nodes based on demand. This intelligent autoscaling system balances the load between chat and source document ingestion to optimize throughput. Administrative parameters allow you to set minimum and maximum limits, as well as throughput thresholds, ensuring efficient resource management and performance. **Q: What are Guardrails?** A: Administrators can easily set up custom guardrails to detect, quantify, and mitigate potential platform abuse. These take the form of both input guardrails, which target unwanted user prompts before they get to the LLM, and output guardrails, which control the LLM's response. **Q: How does Legion handle Hallucinations?** A: Legion leverages Retrieval-Augmented Generation (RAG) to ensure responses are grounded in the customer's data, rather than solely generated by the model's knowledge. If relevant data cannot be retrieved from the customer's environment to answer a prompt, Legion detects this and alerts the user. Additionally, Legion has a hallucination detection system to ensure the model's response is accurate to customer data. If users wish to generate "creative" responses, they can choose to exclude their organization's data sources when interacting with the model. **Q: What is Hallucination Detection?** A: Legion Hallucination Detection feature helps users determine when a response should be verified. Legion highlights areas of uncertainty by marking entities with conflicting information, ensuring responses are highly accurate by clearly indicating when they are not linked to a primary source. **Q: What data types or file formats does Legion support?** A: Currently, Legion supports file formats: .doc, .docx, .htm, .html, .m4a, .markdown, .md, .mp3, .mp4, .pdf, .ppt, .pptx, .rtf, .txt, and .wav. We're constantly working to support new data formats, so this list will grow and evolve over time. **Q: Is Legion mobile or tablet friendly?** A: Yes, Legion is both mobile and tablet friendly. **Q: How does your integration with Confluence work? Are you pre-pulling all data and making a vector database, or is this done on-demand?** A: Legion's integration with Confluence is designed to map data from external sources to a Legion collection. This process involves setting up an integration and defining a cadence for how often new information is ingested into the collection. For example, if the integration is set to refresh every hour, the Legion integration service will check for differences in the data source at the top of each hour. If any differences are detected, only the updated data or documents will be re-ingested to replace the stale data in the Legion collection. This ensures that the data remains current and relevant without unnecessary duplication. **Q: Can Legion function as an OpenAI compliant solution?** A: Legion's Quasar service provides an OpenAI-compatible API on top of the models you choose to host with our platform. This API can be leveraged for various uses and applications, ensuring compatibility with OpenAI standards and practices. **Q: Can we white-label or change the colors of the Legion platform to match our branding?** A: Yes. Legion and its assistants can be customized to match your branding. **Q: Can Legion use Amazon Bedrock or another model server?** A: Yes, Legion's Quasar service can proxy to other model servers, including Amazon Bedrock, if you wish to use them. This flexibility allows you to integrate with various model servers based on your preferences and requirements. **Q: Can I embed your product into my own product/application?** A: Yes, you can embed our product into your own application, and we have experience doing this. If you want to integrate Generative AI capabilities into your product, you can use various components of our platform. For example, you can embed an Assistant directly into your product's interface and white-label it to make it appear as a native part of your application. **Q: Is Legion HIPAA compliant?** A: Yes, Legion is HIPAA Compliant. **Q: Is Legion GDPR compliant?** A: Yes, our product is designed to meet GDPR requirements as needed. We ensure that our data handling practices comply with the regulations set out in the EU General Data Protection Regulation (GDPR) to protect your data privacy and security. **Q: Is Legion SOC2 compliant?** A: Legion is SOC 1 Type 2 and SOC 2 Type 1 & 2 compliant, meeting the Trust Services Criteria for Security, Availability, and Confidentiality. **Q: Is Legion CMMC 2.0 compliant?** A: Yes, Legion is CMMC 2.0 Level 1 compliant. **Q: Is Legion FEDRAMP compliant?** A: Though Legion meets many of the FEDRAMP compliance requirements, we have not pursued official FEDRAMP accreditation. **Q: How does your product ensure that sensitive or Controlled Unclassified Information (CUI) data is not accidentally sent out during Internet searches? How does it prevent accidental exposure when querying internal collections?** A: Legion includes robust access controls for collections to safeguard sensitive data. Internal and Internet calls are kept separate, ensuring that internal data is not exposed to the internet. Additionally, we can display a banner indicating CUI data to raise awareness and prevent accidental exposure. These measures help maintain the security and confidentiality of your data. **Q: There are already open-source models and tools available. Why shouldn't I just build my own LLM for company use?** A: There are several points to consider when deciding whether to buy or build: - **Performance and Accuracy**: Implementing Retrieval-Augmented Generation (RAG) effectively is complex. Our platform maintains high performance and accuracy even with large document sets. We can provide datasets to compare other models' RAG performance against ours. - **Innovation and Support**: We are constantly innovating and enhancing our platform, which our customers benefit from. Building internally would require continuous engagement from your team to keep up with these innovations. - **Enterprise Features**: We have built enterprise features like access controls from the ground up, ensuring they are not just bolt-on solutions. These features are critically important and might be overlooked when building in-house. - **Cost and Efficiency**: Our solution offers a cost-effective approach with a low cost per query, which can be more efficient than building and maintaining an in-house solution. We also provide modular architecture and integration capabilities that can be tailored to your specific needs. **Q: We have our own search capabilities and UI. Is it possible to use just your LLM service to connect to our search system?** A: Because Legion is a modular ecosystem, users are able to plug-and-play their own components into the Legion platform. This means that users can pick and choose which parts of the Legion platform they want to use while continuing to use their preferred internal search, UI, retrieval, or other system. **Q: How can I embed a Legion assistant into another application?** A: After creating an Assistant in our Platform UI, you can take the "embed" code found in the assistant settings and insert it into another application. If you want to place an Assistant or Floating Action Button into an internal website, you can use the relevant code snippets provided to do so. This allows for seamless integration of our Assistant into your existing systems. **Q: Is Legion able to integrate with my custom in-house data storage solution?** A: Yes, our team can create a data integration for nearly any data source. The only requirement is that the data source has a RESTful API that can be used to query the data and ingest it into our platform. This flexibility allows us to seamlessly integrate with your custom in-house data storage solution. **Q: How does access control and data governance work? Is it secure and can it be controlled by the customer?** A: Legion provides robust access control and data governance through role-based access control (RBAC), SSO, MFA, and granular administrative controls. It maintains data sovereignty as a middle-man between your data and LLMs, ensuring information never leaves its environment. It is deployable in secure environments including IL-2, IL-4/5, and IL-6, giving administrators full control over documents, agents, and models. **Q: Are you grabbing data at the time of the chat request or done in batch at a different time?** A: Data ingestion intervals are configurable and managed by your system admin at intervals they choose, leveraging the RESTful API. **Q: How can users get back to the source documents in different systems?** A: Users can get access to the sources used to formulate a response using the sources panel. There they can also view a summarization of the source content. **Q: What is retrieval-augmented generation (RAG), and how does it work?** A: Retrieval-Augmented Generation (RAG) combines a language model with a search mechanism. It retrieves relevant information from a database or documents to enhance the model's responses. This ensures accurate, context-aware answers by using up-to-date data, rather than relying solely on pre-existing knowledge. Learn more: [Medium RAG Article](https://medium.com/enterprise-rag/a-first-intro-to-complex-rag-retrieval-augmented-generation-a8624d70090f) **Q: How does the platform ensure the accuracy and reliability of the information it provides?** A: The Legion platform provides comprehensive citations and source information for all responses. This transparency allows users to see where their answers come from and make informed decisions about the relevance and validity of the data. **Q: Does Legion send any information back to model providers?** A: Legion's approach to sending information back to LLM providers varies on your deployment: - **Legion’s Approach with Proprietary LLMs**: Legion processes, indexes, and searches your data within your environment, only sending the minimal required data to the proprietary cloud-based LLM of your choice (Claude or GPT-4, for instance) to formulate the answer to your question. The LLM does not have access to search, index, or train on your data directly, minimizing the data that leaves your environment. - **Legion’s Approach with Open-Source LLMs**: For maximum security, you can use Legion with an open-source LLM (e.g., Mistral or Llama). This option ensures that your data is fully processed within your environment and your data will never leave your four walls. Our most security-conscious users choose this approach for their deployment for a truly air-gapped solution. **Q: Can I use the Legion platform for classified or sensitive tasks?** A: Yes, Legion is approved for Classified SECRET information and can be deployed in IL-2, IL-4/5, and IL-6 environments. It maintains data sovereignty, ensuring your information never leaves its place and is never used to train AI models. **Q: Are there guidelines or best practices for using the platform effectively? Is any training available to help me become proficient in using the Legion platform?** A: For the latest information, including step-by-step guides on how to use the platform effectively, please visit the help center within your deployment. The help center provides comprehensive resources to ensure you get the most out of the platform. **Q: How often is the data in the platform updated, and how are outdated data handled?** A: The platform typically follows a monthly release cycle, but your system administrator can determine the best time to update or upgrade to the latest version. Data hygiene is managed by your system administrator and users. We recommend regularly purging old or outdated data to maintain optimal performance and results. **Q: Are there limits on the number or type of queries I can perform?** A: With Legion, your system administrator can manage and prioritize resource usage, including setting limits on the number of queries that can be performed. This ensures optimal performance and resource allocation. **Q: Which specific LLMs are used in our deployment, and why were those chosen?** A: The selection and management of LLMs are handled by your system administrator. For detailed information, please reach out to them directly. **Q: What is the process for updating or training the LLMs with new data?** A: The Legion platform does not update or train LLMs with your data. Instead, it segments your data to enable the LLM to query it when generating responses, ensuring transparent and verifiable answers. **Q: Does Legion use any foreign-built models?** A: While SOFchat may include foreign models, Legion conducts rigorous vetting on all models used and deploys them in CVE-free hardened containers to maximize security. **Q: How does the system handle ambiguous or unclear queries from end-users?** A: Clear and specific prompts are essential for generating the best responses. Users should provide as much detail and context as possible in their queries. For guidance on creating effective prompts, please refer to the help center. In cases where users need assistance in providing sufficient detail, Legion assistants can be created to guide users on the information needed to achieve optimal results. **Q: What are the fallback mechanisms if the LLMs fail to generate an appropriate response?** A: If the platform cannot generate an appropriate response based on available data, it will notify you that it cannot determine a response for your query. It may ask for clarification to help provide a related response or offer information based on the data it has access to. **Q: How can users provide feedback on the responses generated by the platform to improve its accuracy?** A: If enabled by the administrator, users can provide feedback using the thumbs up/thumbs down icons within the platform. They can also add detailed comments to explain their rating for a particular response. **Q: Who do I contact for technical support if I encounter issues with Legion?** A: For technical support, please contact support@yurts.ai / support@legionintel.com. The terms of your SLA are included in your contract. **Q: Does Legion have its own coding models?** A: Legion does not provide frontier models and instead leverages third-party, open source, or custom models that the customer provides. There are numerous models that provide advanced support for coding assistance that Legion supports. **Q: How involved do users have to be to set up integrations with Legion?** A: Setting up integrations with Legion is simple and straightforward but requires administrative access and knowledge of the data sources being integrated. It typically involves an administrator accessing the integrations page, configuring authentication details, and setting up mappings into collections. **Q: What should I do if I find incorrect or biased information generated by Legion?** A: If you encounter incorrect information, please use the feedback button (Thumbs up / Thumbs down) within the Legion platform to report it. This will allow us to flag the issue and gather details about the expected response. Regarding biased information, if the data provided by your organization contains inaccuracies or biases, it is important to address these issues internally to ensure the integrity of the information used by Legion. **Q: What is the expected response time for queries submitted to the platform?** A: Response times can vary significantly depending on a multitude of factors, including compute allocation, the type of query (e.g., aggregating or summarizing a large set of documents vs retrieving a specific fact), as well as other environmental variables such as the resources made available to the platform and network latency. Under ideal conditions, responses should begin generating within a few seconds. **Q: What capacity does the platform have in terms of concurrent users or query load?** A: The capacity for concurrent users and query load depends on your deployment, environment, compute bandwidth, and any ongoing background actions. Legion works directly with each customer to ensure the appropriate hardware and resources are secured to efficiently meet user capacity needs. Capacity can be scaled up or down based on the changing or evolving needs of the business. **Q: What can I do with Legion?** A: The possibilities with Legion are significant, but three basic use cases across various industries are: - **Research and Data Analysis**: Find reliable, trusted answers to questions by leveraging your organizational data, complete with comprehensive citations and sources for verifiable information. - **Content Generation**: Use natural language prompts along with your organizational data to create emails, outlines, reports, and more. - **Workflow Automation**: Automate workflows by utilizing your organizational data and Legion's content creation capabilities. Examples include summarizing project details on a specific day and emailing them to your supervisor, or using meeting notes to schedule follow-up meetings with attendees. The possibilities are nearly limitless. **Q: What does Legion do?** A: Legion is a centralized AI platform for securely deploying Generative AI in any organization. Teams that use Legion become more efficient because Legion makes it easy for them to find information, create new content, and automate common workflows. **Q: How accurate is Legion compared to other Generative AI options?** A: Legion uses its proprietary Retrieval Augmented Generation (RAG) in conjunction with the latest models available to deliver best-in-class accuracy. Legion has performed significant testing, including the recognized “needle in a hackstack” (NIAH) benchmark and “multi-needle in a haystack,” in which it has outperformed even the most well-known models. You can see the research for yourself here. **Q: How is Legion different from ChatGPT?** A: ChatGPT connects to the internet. It’s great for brainstorming, quickly getting answers, and small tasks like rewriting something you wrote to sound better. Legion is similar to ChatGPT in that you can use AI to get answers and write new content, but it’s much more. Not only can Legion connect with publicly available data on the internet, but it also connects to all of your organization’s internal data. With Legion, you can do anything from getting a summary of what happened at a meeting you missed, to finding the answer to a problem that someone who left the company solved years ago, or automating the first draft of a report you need to write. Plus, it’s secure enough that even the most skeptical security teams will give it the thumbs up. **Q: How does Legion help find information from different sources?** A: Legion seamlessly integrates with virtually all organizational data sources, using retrieval augmented generation (RAG) to unify information access—allowing users to find critical intelligence across systems without needing to know its specific location, a capability essential for federal knowledge management and decision support. **Q: Which roles typically use Legion day-to-day?** A: Legion is designed to enhance the productivity of any role within an organization. Individuals involved in operations, data management, logistics, decision-making, after-sales service teams, and project coordinators all can leverage Legion’s platform to find information faster, generate new content, or automate their workflows. **Q: What measures are in place to control and monitor access to my data?** A: Legion provides comprehensive access control and monitoring through: - Role-based access control (RBAC) for secure user management - Single Sign-On (SSO) and Multi-Factor Authentication (MFA) for secure access - Granular administrative controls at document, agent, and model levels - Permission-based access by role, group, or individual user - Real-time metrics and performance dashboards for monitoring activity Legion's architecture gives administrators complete visibility into resource allocation, prioritization, and compute management. The platform ensures users only access data they're authorized to see, maintaining strict data sovereignty while providing administrators the tools to monitor and control all system interactions. These controls operate within Legion's secure middle-man architecture that prevents any data from leaving your environment or being used to train external models. **Q: How do you collect, store and use the data input into the Legion platform?** A: Legion only collects the minimum data necessary to provide our AI-powered services. This typically includes the queries and documents you input for analysis. We do not collect any extraneous user data. All collected data is securely stored and used solely to process your requests and improve our models, never for any other purpose. **Q: What are your data retention and deletion policies?** A: Your data belongs to you and is retained only as long as your account is active or as required by law. You can request complete deletion of your data at any time, which will be processed within 48 hours. Routine deletion of inactive customer data occurs every 60 days. **Q: Who owns the IP of the outputs generated by Legion AI?** A: You maintain full ownership of any data you input into our system as well as the outputs generated using that data. We claim no intellectual property rights over your data or the insights and content our AI generates for you. Your data is not used to train our general models without express consent. **Q: What third parties, if any, have access to the data I put into the Legion platform?** A: Your data is never shared with any third parties without your explicit permission. The only exception is if you opt to use third-party AI models or APIs in conjunction with our platform. In that case, the data is shared only to the extent necessary to process your request. **Q: How can I monitor and audit the way my data is being used?** A: The Legion platform includes built-in tools for data governance and auditing. You can view detailed logs of all data access and usage through your admin dashboard. We can also provide periodic usage reports and work with your team to set up automated monitoring and alerts. **Q: What relevant security and privacy certifications does Legion maintain?** A: Legion is GDPR, HIPAA, SOC 1 Type 2, SOC 2 Type 1 & 2, and CMMC 2.0 compliant, and has been approved for use in secure environments IL-2 through IL-6 (Classified SECRET). Legion is trusted by the US Department of Defense (DoD) and Department of Energy (DoE), demonstrating its ability to meet high security standards. Legion's development practices leverage DISA strict security guidelines to create highly secure software that meets stringent security requirements, which helps shorten approval times for government deployments. **Q: How can generative AI improve diagnostics and repairs processes in MRO operations?** A: Generative AI can significantly enhance diagnostics and repair processes by analyzing observed issues, quickly identifying probable causes, and providing repair guidance. It can integrate with existing manuals and standard operating procedures (SOPs) to deliver precise, actionable instructions, reducing downtime and improving repair accuracy. **Q: In what ways can Generative AI assist MROs in adhering to regulatory compliance?** A: Generative AI helps MROs manage Airworthiness Directives (ADs) by tracking status, managing compliance due dates, generating reports, and providing maintenance solutions. This ensures adherence to regulations from authorities like the FAA and EASA, preventing operational and brand damage. **Q: How does Legion differ from Large Language Models (LLMs) like Chat GPT or Claude, specifically for MRO use-cases?** A: Legion is tailored for aerospace MRO environments, delivering exceptional security and precision. Unlike standalone LLMs such as ChatGPT or Claude, Legion safeguards data by serving as a secure intermediary between the LLM and your data. Supporting multiple LLMs (both open-source and proprietary) for maximum flexibility and security, Legion can operate fully on-premises, even in air-gapped setups. It integrates seamlessly with existing MRO systems to provide actionable intelligence based on technical documentation. **Q: How is compute charged?** A: Legion is designed to enhance your existing systems. It does this by connecting to all of your information, whether it lives in an MES, database, or something else, and aggregating it into one easy-to-use chat interface. Best of all, that chat interface can be embedded seamlessly into any of your pre-existing systems. Users won’t even realize they’re using a new tool - they’ll just know their workflow got far more efficient. **Q: We already use AWS. Why wouldn't we just use Bedrock as our easy button for GenAI?** A: AWS Bedrock is a foundational service, it gives you access to LLMs and basic tools to build GenAI applications. But Bedrock is not a complete enterprise-grade solution. Legion builds on top of these foundations to offer a full-stack, secure, and turnkey platform designed specifically for enterprise and government-grade AI operations. **Q: What does Legion provide that Bedrock does not?** A: Legion goes far beyond just serving models. It offers a full enterprise-ready platform with capabilities that Bedrock simply doesn’t include out of the box: - A user-facing chat UI for secure, structured GenAI interaction — no need to build one yourself - Role-based access controls and permissions to protect sensitive data and manage usage - SSO and identity provider integrations (Okta, Azure AD, etc.) for seamless authentication - Pre-built integrations with your existing data sources — connect to CRMs, data lakes, SharePoint, and more without custom dev work - Admin console with fine-grained feature controls, user auditing, and policy enforcement - Zero-trust architecture, baked into the platform from the ground up - Cloud-agnostic and portable deployments, supporting AWS, Azure, on-prem, and edge environments with a consistent experience - Compliance-ready foundation, with privacy, logging, and security built in Bedrock is great if you want raw model access. Legion is what you need when you're ready to actually run AI in production across your enterprise. **Q: Can't we just build the missing parts ourselves on Bedrock?** A: Technically yes—but at significant cost. You’d be committing to building and maintaining: - A frontend UI for secure GenAI usage - Access control & permissions system - Model orchestration logic - Data integrations - Enterprise auth & RBAC features - Logging, observability, and user auditing - Custom policies for data handling, privacy, and governance With Legion, all of that is already built, hardened, and production-tested. You're getting a complete AI workflow system, not a dev toolkit. **Q: Can Legion work with Bedrock under the hood?** A: Yes! Legion can integrate directly with Bedrock, using its hosted LLMs. You don’t lose your investment in AWS—you extend it with enterprise-grade tooling. Think of Legion as your intelligent GenAI control plane that can abstract across Bedrock, Azure OpenAI, Anthropic, open-source models, and more. **Q: What if we need flexibility across clouds or for regulated environments?** A: That’s where Legion shines: - **Cloud-Agnostic**: Deploy in AWS, Azure, GCP, or hybrid environments - **On-Prem & Edge-Ready**: Ideal for disconnected or sensitive environments - **Consistent UX**: Same enterprise experience across all deployments - **Data Sovereignty**: Your data stays where you want it—always under your control **Q: Who should use Legion?** A: - Enterprises that need security, governance, and compliance - Government agencies requiring on-premise or edge deployments - Organizations that want to move faster than custom development cycles - Teams that want to focus on insights, not infrastructure **Q: What is an “Agentic AI Workflow”?** A: It’s a secure, task-specific AI agent that performs multi-step processes autonomously using your internal systems—gathering data, executing logic, and generating results without ever leaving your network, while ensuring full auditability so you can track and review every action taken by the system. **Q: How does Legion handle data versioning and document updates?** A: Legion supports time-aware retrieval and dynamic context updates to ensure models always reflect the latest version of your data sources, with full traceability. **Q: What kind of infrastructure is required to run Legion?** A:Legion runs efficiently with minimal GPU overhead (2 GPUs for many deployments). For larger, more intensive workflows, multi-GPU setups can be employed, with compatibility across cloud or bare-metal environments.