AI Driven Image Search Tool

The United States Patent and Trademark Office (USPTO) issued a Request for Information (RFI) to explore commercial Software as a Service (SaaS) solutions for an AI-driven image search tool. This tool will aid in processing and granting patents. The RFI, facilitated by the Office for the Chief Financial Officer, Office of Procurement, is a preliminary market research effort to identify qualified vendors for an alternative competition method under the Patent and Trademark Office Acquisition Guidelines (PTAG). The process involves two phases: a Vendor Questionnaire (Phase 1) due by June 19, 2026, and Product Demonstrations and Interactive Dialogue (Phase 2), anticipated between July 13-17, 2026, for selected vendors. Only vendors responding to the RFI will be considered for the alternative competition. The USPTO aims for a complete turnkey solution for all three patent types and prefers FedRamp-finalized solutions, though FedRamp in-process solutions may be considered.

This government Request for Proposal (RFP) outlines requirements for a Software-as-a-Service (SaaS) solution to augment the United States Patent and Trademark Office's (USPTO) existing enterprise systems. Key requirements include FedRAMP authorization, with a detailed description of continuous monitoring, Plan of Action and Milestones (POA&M) management, and Security Impact Analyses (SIA) for system updates. The RFP also seeks technical capabilities for customizing UI/UX workflows, specifically integrating with Okta (OpenID Connect) and implementing automated export/redaction builders. Furthermore, the solution must demonstrate a technical approach to ingesting, normalizing, and maintaining Global Design, Utility, and Plant patent data weekly, with an AI architecture capable of processing complex, multi-image queries (e.g., auto-lasso) to deliver accurate similarity rankings and CPC/Locarno classification suggestions within a sub-6.0 second search execution Service Level Agreement (SLA). Vendors may propose solutions supporting any portion of these patent types.

The United States Patent and Trademark Office (USPTO) seeks a FedRAMP-authorized, AI-driven Software-as-a-Service (SaaS) image search tool to enhance prior art search capabilities for its 12,000 Patent Examiners. This tool will augment existing systems by allowing examiners to use images as direct queries in global patent databases, improving search completeness and efficiency, especially for visual design and utility applications. The scope includes providing, configuring, customizing, and maintaining the platform, which must aggregate global patent and trademark databases from at least 80 international authorities. Key features required are robust AI-powered image-to-image search, multi-image uploads, image manipulation tools, similarity ranking, and automated classification suggestions (USPC, Locarno, CPC). The SaaS solution needs extensive customization for user interfaces, examiner workflows, and data export reports to align with USPTO practices. The contractor will be responsible for project management, ensuring security, privacy, and AI transparency in compliance with federal mandates like FISMA and NIST guidelines, and integrating with USPTO’s Identity Management via Okta. Training will utilize a Train-the-Trainer methodology for USPTO personnel, including super-user training and providing comprehensive trainer kits and ongoing support. The contract specifies intellectual property rights, with the USPTO retaining ownership of all customized code and government data, while the contractor retains rights to the base SaaS platform.

The document outlines a structured format for collecting and categorizing questions and answers, likely within a government procurement or grant application process. It includes fields for 'Question #', 'Reference', 'Question', 'Answer', and 'Question Category (Contract or Technical)'. This format enables clear organization and classification of inquiries, distinguishing between contractual and technical aspects. Such a system is essential for maintaining transparency, ensuring consistent responses, and facilitating efficient communication between an entity (e.g., a federal agency issuing an RFP or grant) and potential contractors or grantees. The structure suggests a formal Q&A process, crucial for clarifying requirements, addressing ambiguities, and supporting informed decision-making throughout the lifecycle of federal RFPs, federal grants, or state and local RFPs.

This government file outlines a comprehensive set of deliverables across five key tasks: Project Management, Security & AI Compliance, SaaS Customization (Special Works), Database & Search, and Training. The deliverables span from initial transition plans and ongoing reporting to technical specifications like FedRAMP authorization, AI transparency documentation, and customized SaaS portal development. Key aspects include security assessments, system configurations, database updates, and comprehensive training programs, all with defined due dates and frequencies. This structured approach ensures meticulous oversight, adherence to security protocols, and effective implementation and user adoption of the system.

This government file outlines the performance objectives, associated deliverables, performance standards, acceptable quality levels (AQLs), surveillance methods, and incentive/disincentive considerations for a government contract. The key objectives include project management and transition, system availability and performance for a SaaS platform, security, privacy, and AI compliance (including FedRAMP ATO and secure data handling), SaaS customization and deployment (with Okta integration and automated redaction), database aggregation and search accuracy for global patent/design databases, and