Executive Summary
Hawaii law (HRS §27-43) charges the Office of Enterprise Technology Services (ETS) and its Chief Information Officer with statewide IT governance. Under the Governor’s Budget Execution Policy (EM-25-03, Section 20.d), every IT acquisition of $100,000 or more—hardware, software, projects, maintenance contracts, consultant services, and more—requires CIO approval. Departments submit an IT Spend Request through the state’s IT portfolio management system with supporting documentation. Projects at $1 million or above, or those designated by the CIO as enterprise initiatives, must additionally be reviewed and approved by the Project Advisory Council (PAC), whose voting members are the CIO, the DAGS Comptroller, the Director of Finance, and the Director of Human Resources Development (Administrative Directive 18-03).
This governance structure generates hundreds of IT Spend Requests each fiscal year. Every request requires evaluation across multiple dimensions—cost justification, strategic alignment with state IT goals and the department’s multi-year IT roadmap, execution risk, impact of inaction, vendor history, and comparison with similar past initiatives. Until now, this evaluation has been entirely manual: analysts cross-reference data from the state’s enterprise architecture platform, research vendors and market pricing independently, and produce narrative assessments without a standardized framework. The result was inconsistent evaluations, slow turnaround, and no institutional memory across assessment cycles.
Hawaii built an AI-powered assessment system that transforms this process. When a reviewer triggers an assessment, the system automatically retrieves the complete project record from the state’s enterprise architecture platform, searches a knowledge base of over 1,400 historical IT Spend Requests for relevant precedents, researches current market pricing and vendor government contract history via web search, and synthesizes all findings into a comprehensive Word document assessment—delivered in two to four minutes.
Since deployment, the system has reduced assessment preparation time from hours of manual research to minutes of automated analysis, established consistent evaluation standards across all requests, and created institutional memory that improves with every project added to the knowledge base. The architecture is entirely cloud-native, using serverless computing, and the approach is replicable by any state with an enterprise architecture platform and a body of historical project data.
