GRACE AI - SECURE GENERATIVE AI ASSISTANT

Private FedRAMP-Compliant Research Acceleration Platform

  • Advanced Research Projects Agency for Health (ARPA-H)

  •  ITSS (Information Technology Support Services)

  • 75N92023F00253

  • $200,000

  • 2024-2025

  • Time & Materials (T&M) and Firm Fixed Price (FFP)

  • Prime Contractor (delivered through Phase No. 5)

Challenge

ARPA-H researchers and program managers needed access to advanced generative AI capabilities to accelerate biomedical research, analyze scientific literature, and synthesize complex health data. However, using commercial AI platforms like ChatGPT posed significant security risks including potential data exposure, lack of federal compliance, inability to access classified or sensitive research data, and no integration with agency-specific knowledge repositories. The agency required a secure, compliant AI solution that could enhance productivity while maintaining the highest standards of data protection and federal information security.

Solution

DAWNE IT Solutions (through Phase No. 5) designed and deployed GRACE (General Research And Content Engine), a secure, private generative AI assistant tailored specifically for ARPA-H's research mission. Named in homage to pioneering computer scientist Grace Hopper, the platform delivered ChatGPT-4 level capabilities within a FedRAMP-compliant environment with enhanced access to biomedical research databases and ARPA-H internal knowledge.

Core Capabilities

    • Dedicated, isolated AI instance for ARPA-H exclusive use

    • GPT-4 level natural language processing and generation

    • Conversational interface familiar to commercial AI users

    • Multi-turn dialogue with context retention

    • Code generation and technical documentation support

    • Research synthesis and summarization capabilities

    • Real-time inventory visibility across all ARPA-H locations

    • Automated tracking of equipment assignments and transfers

    • Government property identification decal management

    • Peripheral device tracking (monitors, keyboards, docking stations, etc.)

    • Office supply inventory monitoring and replenishment

    • Equipment pooling for temporary assignments

    • PubMed Integration: Direct access to 35+ million biomedical research citations

      • Automated literature search and retrieval

      • Citation extraction and bibliography generation

      • Research trend analysis across publications

      • Author and institution network mapping

    • HHS Data Access: Integration with Department of Health and Human Services datasets

      • Public health statistics and epidemiological data

      • Clinical trial registries and outcomes

      • Healthcare policy documents and regulations

      • Federal health program data

    • ARPA-H Internal Documents: Secure access to agency knowledge repositories

      • Internal research reports and findings

      • Program documentation and proposals

      • Meeting notes and decision records

      • Strategic planning documents

      • Policy and procedure manuals

    • Rapid literature review and synthesis

    • Research hypothesis generation

    • Experimental design recommendations

    • Data analysis methodology suggestions

    • Grant proposal drafting assistance

    • Scientific writing and editing support

    • Cross-disciplinary connection identification

    • Research gap analysis

    • Private cloud deployment with data isolation

    • FedRAMP-compliant infrastructure

    • End-to-end encryption for all queries and responses

    • No data sharing with external AI providers

    • Comprehensive audit logging of all interactions

    • Role-based access controls

    • Multi-factor authentication

    • Data residency within federal boundaries

    • Document upload and analysis

    • Batch processing for multiple documents

    • Export capabilities for research outputs

    • Collaboration features for team projects

    • Version control for generated content

    • Integration with Microsoft 365 environment

Technologies Used

AI Platform

GPT-4 architecture (private deployment)

Security

Multi-factor authentication, encryption at rest and in transit

Collaboration

Microsoft Teams integration for shared research

Cloud Infrastructure

FedRAMP-compliant cloud environment

Access Management

Okta SSO integration

Data Integration

PubMed API, HHS data portals, SharePoint

Monitoring

Comprehensive logging and audit trail systems

Implementation Phases

Phase 1: Architecture & Security Design

  • FedRAMP compliance assessment and planning

  • Security architecture design and documentation

  • Data classification and handling procedures

  • Access control policy development

  • Integration architecture planning

Phase 3: Testing & Validation

  • Security penetration testing

  • Functionality validation across use cases

  • Performance optimization

  • User acceptance testing with research teams

  • Compliance verification and ATO preparation

Phase 2: Platform Deployment

  • Private AI environment provisioning

  • Security controls implementation

  • PubMed API integration development

  • HHS data source connections

  • ARPA-H document repository integration

  • User authentication and authorization setup

Phase 4: Training & Rollout

  • User training for researchers and program managers

  • Best practices documentation

  • Use case examples and templates

  • Responsible AI usage guidelines

  • Ongoing support procedures

Results & Outcomes

  • 60% reduction in literature review time for research teams

  • 45% increase in researcher productivity for grant proposal development

  • 100% data security compliance with zero security incidents

  • 85% user satisfaction rating among ARPA-H researchers

  • Accelerated research hypothesis generation and validation

  • Enhanced cross-disciplinary research connections

  • Improved data-driven innovation capabilities

  • Strengthened health ecosystem connectivity

  • Facilitated successful research transitions from concept to implementation

  • Reduced risk of sensitive information exposure compared to commercial AI platforms

Compliance & Standards

  • Federal Risk and Authorization Management Program compliance

  • Federal Information Security Management Act compliance

  • Security and privacy controls implementation

  • Full authorization maintained

  • Compliance with federal data handling requirements

  • Ethical AI usage guidelines and monitoring

Use Case Examples

  • Researchers used GRACE to synthesize hundreds of research papers on emerging biomedical topics, reducing weeks of manual review to hours of AI-assisted analysis.

  • Program managers leveraged GRACE to draft sections of research proposals, identify relevant prior work, and ensure comprehensive literature citations.

  • Scientists used GRACE to explore cross-disciplinary connections and generate novel research hypotheses based on existing literature and data patterns.

  • Research teams consulted GRACE for methodology recommendations and statistical approach validation for complex biomedical datasets.

  • Staff used GRACE to quickly locate and interpret relevant HHS regulations and federal health policies applicable to research programs.

Key Personnel

Innovation & Impact

  • GRACE represented one of the first secure, mission-specific generative AI deployments in the federal health research space, establishing a model for responsible AI adoption in government.

  • By providing secure AI capabilities, GRACE directly supported ARPA-H's mission to accelerate biomedical breakthroughs and transform health outcomes through advanced technology.

  • The success of GRACE AI inspired DAWNE IT Solutions to develop the Government Employee AI Mastery Program (GEAMP), a comprehensive training initiative to help federal employees across all agencies leverage AI technologies effectively and securely.

  • The project established best practices for federal AI deployment that can be replicated across other agencies seeking to adopt generative AI while maintaining security and compliance.

Previous

ENTERPRISE ASSET MANAGEMENT SYSTEM IMPLEMENTATION
Next

COMPREHENSIVE IT SUPPORT SERVICES & DESKTOP-AS-A-SERVICE MIGRATION