AI Development & Implementation - Request for Proposals (RFP)

Learn more and apply here: https://eradicatehate.org/about/opportunities/ Statement of Intent Eradicate Hate is seeking a skilled AI development and implementation partner to build, deploy, and maintain the Prevention Resource Assistant across web properties.  Our goal is to offer short, guided chat experiences that route prevention practitioners, individuals, and families to trusted and tailored resources in a unified prevention resource knowledge base.   We are particularly interested in proposals that address customization for the various audiences we support and strong protections for user safety in sensitive conversations.  Strong proposals will not just address the AI backend but also include a simple web interface that can easily integrate into our various web properties. The selected partner will demonstrate technical expertise, a strong understanding of safe and ethical AI design, experience delivering mission-aligned tools for nonprofits, and a deep understanding of our organization and the violence prevention space.  The partnership will include both system buildout and ongoing engagement with Eradicate Hate’s content and operations team for organizational adoption, issue management, and feature development.  Proposals can include multiple contractors with an emphasis on partner collaboration, both existing and new partners. Project Goals and Scope of Services A. System Architecture and Development • System: Develop a scalable, cloud-based architecture supporting a shared AI knowledge base with multiple role-specific chat interfaces (e.g., parents, students, practitioners). • AI: Implement a chatbot that puts users in touch with helpful resources through short chats with highly accurate references to knowledge base documents. • Integration: Embed the widget seamlessly across web properties while using a unified backend. • Repository & Citation System: Implement a searchable repository of documents (PDFs, video transcripts, webpages) and a citation system for the AI to refer users to the appropriate link or resource.  The repository should scale to accommodate and organize thousands of documents for ongoing content management. • Guardrails & Safety: Integrate robust safety guardrails, including content gating by audience, static crisis resource cards, prompt injection mitigations, and AI output monitoring/interception.  Checks for correspondence between model output and actual source document text.  Implement per-version red-teaming for safety against prompt injection, misinformation, and sensitive conversations. • Security & Resiliency: Demonstrate best practices in infrastructure security and secure development, including encryption of data in transit and at rest; endpoint safeguards for DDoS, spoofing, and injection attacks; developer MFA and regular credential rotation; support for multi-region deployment and failover, and multi-stage CI/CD pipelines with role-based access controls and code review enforcement. Proposals should address monitoring, incident response, and alignment with relevant nonprofit or government security frameworks (e.g. NIST CSF or equivalent cloud security standards). • Localization & Accessibility: Design for ADA compliance, multilingual expansion, inclusive UX. B. Knowledge Base & Content Workflow • Publisher Workflow: Build a light editorial interface for Eradicate Hate and partners to upload, tag, and update documents, including automated text extraction and video transcription, and human review workflows. • Version Control: Maintain version history and vetting status to ensure reliability of displayed content. • Data Protection: Design for strict privacy and data minimization standards, including log anonymization and PII censoring. C. Analytics, Continuous Improvement, and User Support • Analytics: Provide access to KPIs (e.g., time-to-resource, helpfulness ratings, conversation completion rates, missing content) via a simple dashboard or report. • Continuous Improvement: Provide a means of iterating on model instructions, as well as a systematic means for requesting changes or new features. • User Support: Provide SLAs for user support, including expected level of service, process for raising and addressing issues, and target turnaround times. • Partnership: Establish a collaborative relationship with Eradicate Hate and its partners to support product delivery, maintenance, and growth. Budget Work is planned to begin January 10, 2026, and conclude January 31, 2027.  This contract will be firm-fixed price inclusive of all usage; proposals should offer pricing based on a fixed number of short (6-8 message) conversations.  Budgets should be formatted based on upfront development and monthly ongoing costs. Evaluation Criteria Applicants will be selected based on the following desired criteria: ·      Understanding of scope and objectives ·      Exemplar prior work product and results; references are strongly encouraged ·      Proposed approach and methodology ·      Budget and value for money ·      Ability to work as a team with Eradicate Hate and its partners ·      Timeline and project management ·      Creativity of deliverables ·      Strength of partner collaborations, collaborative proposals are welcomed ·      Knowledge of preventing hate-fueled violence Equal Opportunity Statement: Eradicate Hate will make a good-faith effort to identify and solicit minority- and women-owned vendors.  The Eradicate Hate Global Summit shall not discriminate on the basis of race, ethnicity, gender, sexual orientation or disability in its selection of vendors and contractors. Proposal Submission Interested applicants should submit their proposal via email to jobs@eradicatehatesummit.org.  Proposals should include: • Scope of Work: a proposed action plan given the “Statement of Intent” and “Project Goals and Scope of Service” listed above. • Budget: this contract is a firm-fixed price contract; budgets should be formatted based on monthly or quarterly payment structure. • Team Overview: a list of the team members who will be contracting on this project, along with a short description of their roles. Eradicate Hate will selectively interview potential awardees based on an initial review of applications. Anticipated Selection Timeline The Request for Proposal will be open through November 12, 2025.  Eradicate Hate aims to make a selection by November 21, 2025, and will notify candidates after the selection is complete. For questions regarding the RFP, please feel free to reach out to Program Manager of the Prevention Practitioners Network, Neil Saul at nsaul@eradicatehatesummit.org. Learn more and apply here: https://eradicatehate.org/about/opportunities/