Artificial intelligence: A strategy to pay nonprofits on time and relieve staffing pressures

Artificial Intelligence is a hot topic these days. It’s moving markets, underpinning public policy decisions, and redefining the workforce.  Like many technology innovations, AI has been slower to gain traction in the public sector.  In New York City, however, we can no longer afford to delay its adoption. Our city’s $20 billion nonprofit human services contracting portfolio is facing unprecedented economic hardship and sustainability concerns due to late contracts and invoice payments. A frequent refrain from government partners is that the procurement workforce is understaffed and undertrained. AI is a solution that can address both of these issues. AI-powered tools can automate tasks and are designed to quickly learn digital systems, like the city’s electronic procurement platform. 

In a ChatGPT query, I asked: “How can AI be used in human services government procurement?” What follows is the chatbot’s general recommendations. Interviews with New York City procurement staff provided examples of specific applications and potential implications for integrating AI in New York City’s human services contracting portfolio. The city just released a request for proposals to enhance PASSPort functionality – this article urges the city to integrate AI functionality into future enhancements. 

ChatGPT Recommendation: Automating Procurement Processes – RFP Drafting, Evaluation, Review, Compliance & Scoring. AI tools can help agencies generate procurement documents, such as RFPs and RFIs by analyzing past templates. AI-powered tools can review proposals to ensure that they meet legal and policy requirements and to highlight key points for decision makers, speeding up evaluations and reducing human bias. 

In New York City, vendors often submit proposals for human service delivery that exceed 100 pages. Currently, agency procurement professionals must manually review each proposal to determine responsiveness and to evaluate the vendor’s approach to service delivery.  This process can take weeks, if not months. AI can streamline this task by instantly identifying non-responsive proposals (i.e., proposals that are incomplete or do not answer the questions that were asked). For responsive submissions, AI can conduct a comparative analysis and cross-check proposals against RFP requirements – all within minutes and without requiring human resources. While entrusting the entire evaluation process to AI may seem daunting – and likely premature – leveraging AI to reduce workload and pinpoint key areas for deeper review could significantly accelerate contracting.

Chat GPT Recommendation: Supplier & Vendor Management. AI can evaluate a vendor’s financial stability, past performance, and regulatory compliance. 

New York City requires that every vendor under consideration for a contract must be deemed responsible, a determination that can involve reviewing hundreds of pages of information and take weeks to complete. Some responsibility determinations include lengthy court decisions, Department of Investigation memos, multiple years of financial statements, global news reports, and more. AI can digest and analyze this data in minutes, comparing findings against vendor requirements and risk factors.

ChatGPT Recommendation: Contract Management, Fraud Detection & Risk Management.  AI-powered contract analysis tools can track deadlines, flag compliance risks and suggest modifications. Machine learning models can detect anomalies in bidding patterns, pricing or contract awards to prevent fraud and waste. 

In New York City, PASSPort data can identify and manage vendor and contracting risks. With current challenges in extracting reports from the system, AI provides an alternative, user-friendly and more efficient approach allowing for quick retrieval and real-time analyses. For example, an agency procurement professional – or a centralized risk management team - could query PASSPort with the following questions and then make strategic management and policy decisions based on the information – all within minutes:

• “How many vendors are not spending according to budgets?”
• “Which vendors are behind on invoice submissions?”
• “What is the median number of invoice revisions per contract?” 
• Which vendors are above and below the median?”

ChatGPT Recommendation: Chatbots & Virtual Assistants for Vendor Assistance and Internal Procurement Support. AI-powered chatbots can guide vendors through the procurement process, answer common questions and reduce administrative burdens. AI virtual assistants help procurement officers by providing quick access to policies, templates and historical data. 

In New York City, there is often a heated debate about delayed invoice payments. Providers claim agencies are behind on payments, while agencies argue that providers often submit incomplete invoices. PASSPort could be enhanced with AI to resolve this back and forth once and for all.  When a vendor submits an invoice, AI can compare it against the budget, verify completeness, and confirm alignment with required documentation.  Any red flags, discrepancies, or missing information would be immediately identified for the vendor to correct before resubmission, expediting agency approval. Currently, agencies often return invoices for a single error, only to return them again for another mistake. AI would enable a comprehensive review upfront, identifying all issues at one time—a faster, more customer-friendly approach. Additionally, AI queries could determine the source of nonprofit cash flow challenges with the following questions providing a quick pathway to resolution:

• “What percentage of budgets are awaiting invoice submissions?”
• “What percentage of invoices are pending agency approval?”

ChatGPT Recommendation: Enhancing Transparency & Compliance, Automating Contract Management, Audit & Reporting.  AI powered tools can track deadlines, flag compliance risks, and suggest contract modifications.  AI can generate real-time reports on procurement activities, improving accountability and public trust. 

New York City frequently launches “Clear the Backlog” initiatives to expedite contract registration and invoice payments. They require manually creating and updating a centralized procurement dashboard through weekly PASSPort data pulls, validation from more than a dozen agencies and oversights and email and phone communications.  All of this activity is in addition to regular procurement tasks. With AI, agencies could independently establish their own dashboards and manage their portfolios through multi-step queries, such as:

• “List all agency contracts that are late in registration, sorting them from most to least overdue, and identify the procurement process stage where they are stalled.”
• “Analyze the last two years of agency contracts and identify the three tasks that consistently take the longest time.”

Centralized oversight, like the Mayor’s Office of Contract Services and City Hall offices, could also run these queries to identify high performing contracting agencies, those that need support and where contracts are stuck.  This level of automation would empower city officials to monitor procurement efficiency, predict potential bottlenecks, and identify targeted opportunities for policy reforms and/or workforce training. 

New York City’s PASSPort system, powered by Ivalua software, can be enhanced with AI-driven features such as those described above. The city recently released an RFP to upgrade PASSPort, yet AI capabilities were not included.  It’s not too late to amend the procurement by integrating  AI functionality into PASSPort to expedite human services contracting and fast-track timely payments to nonprofits.