Article reader Listen + reading controls
Article reader
Preparing the reader…
Reading settings
AI Talent Needs Durable Product Funding, Not Episodic Enthusiasm¶
The Department of Veterans Affairs’ difficulty hiring artificial-intelligence talent under uncertain appropriations illustrated a structural contradiction in federal modernization. Leaders can recognize AI as a generational capability, identify important health and benefits applications, and receive new governance mandates—while lacking the durable funding needed to assemble and retain the teams responsible for the work.
This is not simply a human-resources problem. Funding instability changes architecture, acquisition, accountability, and risk.
An agency that cannot sustain internal technical capacity becomes dependent on vendors for the knowledge required to understand its own systems. It may still acquire AI. It loses the ability to act as an intelligent owner.
VA leaders have consistently emphasized mission as a recruiting advantage and the need for competitive tools such as special salary rates and public-service pathways. In a November 2023 press conference, VA Secretary Denis McDonough discussed the value of salary flexibility, Presidential Innovation Fellows, the U.S. Digital Service, and the mission of serving veterans. VA later published an AI Workforce Resources Blueprint describing workforce development as a linchpin of responsible AI adoption.1
The blueprint is important. A workforce strategy requires a corresponding funding strategy.
AI Systems Create Continuing Obligations¶
A conventional project budget often assumes a build phase followed by lower-cost operations and maintenance. AI complicates that pattern.
After deployment, teams must continue to:
- monitor data quality and population change;
- evaluate model and system performance;
- investigate incidents and appeals;
- maintain software, infrastructure, and security;
- update knowledge sources and interfaces;
- retrain or replace models where justified;
- communicate limitations to users;
- and reassess whether the system still improves the service.
These are not optional enhancements. They are part of operating a high-consequence system responsibly.
Funding a pilot without funding the product lifecycle creates pressure either to abandon a useful capability or to leave it running without adequate stewardship.
Permanent Capacity Protects Public Accountability¶
VA can and should use contractors, cloud providers, researchers, and commercial AI products. It still needs federal employees with enough expertise to make independent decisions about:
- mission priorities and product scope;
- data authority and use;
- architecture and vendor dependency;
- evaluation sufficiency;
- privacy, civil rights, and clinical or benefits risk;
- authorization of deployment and expansion;
- incident response and suspension;
- and whether a contract is delivering public value.
These are sovereign and fiduciary responsibilities. They cannot be fully outsourced because the contractor’s commercial incentives and the agency’s public obligations are not identical.
Durable internal capacity also preserves institutional memory when vendors change. An agency should not have to relearn why a model, threshold, or workflow exists every time a task order turns over.
Continuing Resolutions Distort the Talent Market¶
Skilled candidates evaluate risk. A temporary appointment, uncertain team budget, delayed equipment, or unclear product future makes public service less competitive even when the mission is compelling.
Funding uncertainty also slows the hiring machinery:
- leaders hesitate to open positions;
- HR cannot make timely commitments;
- managers cannot describe a credible team or roadmap;
- candidates accept other offers;
- existing employees absorb vacancies and burn out;
- and contractors fill urgent gaps under mechanisms that may cost more over time.
The result is a reinforcing loop: unstable funding weakens capacity, weak capacity slows delivery, slow delivery makes long-term investment harder to defend.
Breaking the loop requires funding teams against outcomes rather than waiting for each individual project to justify a temporary position.
Fund Product Teams as the Unit of Capability¶
VA’s AI workforce should be organized around durable mission products and shared enabling services.
A mission product team might own an end-to-end service such as claims-document processing, appointment support, clinical decision support, veteran communications, or fraud detection. The team persists as models and vendors change. It includes product, domain, software, data, ML, security, design, and evaluation capacity appropriate to the risk.
Shared teams can provide:
- secure development and model platforms;
- data and knowledge infrastructure;
- reusable evaluation and red-team services;
- privacy, legal, and responsible-AI patterns;
- observability and incident tooling;
- and acquisition support.
Funding these capabilities once reduces duplication and gives new hires an environment in which they can produce value.
Mission Is an Advantage Only When Work Reaches the Mission¶
Public service can attract people who want their work to matter. That promise erodes when organizational friction prevents delivery.
Retention depends on:
- access to users and domain experts;
- modern tools and representative environments;
- clear product ownership;
- technical career progression;
- leaders who resolve cross-organizational barriers;
- time for learning and quality;
- and visible connection between technical work and veteran outcomes.
The DigitalVA account of partnerships attracting technical talent emphasizes both mission and an environment that supports professional excellence. Compensation narrows the gap; professional agency makes the mission credible.
Workforce Development Must Be Tiered¶
VA’s scale means AI capability cannot reside in a central expert group alone. The workforce needs different levels of competence:
- AI literacy: staff understand approved uses, limitations, privacy, security, and escalation.
- AI-enabled practice: clinicians, benefits professionals, and operational staff use and evaluate tools in their domain.
- Product and governance leadership: managers define outcomes, decision rights, evidence, and risk.
- Technical building and operations: engineers and scientists design, integrate, deploy, and monitor systems.
- Independent evaluation: specialists challenge performance, safety, security, civil-rights, and human-factors claims.
Training should be connected to real work, mentorship, and career paths. A catalog of courses does not substitute for funded positions in which new competence can be exercised.
Budgeting Should Make Lifecycle Costs Visible¶
AI proposals should include a workforce and operating-cost model across several years:
- government and contractor labor;
- data acquisition and stewardship;
- compute, storage, and model access;
- integration and software maintenance;
- evaluation and monitoring;
- security and incident response;
- user training and change management;
- accessibility and appeal;
- and exit or replacement.
Leaders can then compare AI with the baseline service honestly. A pilot that appears inexpensive because long-term stewardship is omitted is not efficient; it is underpriced.
Portfolio funding should also allow agencies to stop weak use cases and move people to stronger ones without dismantling the team. Stable capability funding does not mean permanent commitment to every product.
Measure Institutional Capacity¶
VA should measure more than the number of AI hires or use cases:
- time from hiring to useful contribution;
- critical product roles filled;
- retention and internal mobility;
- percentage of high-impact systems with a named government owner;
- ability to perform independent evaluation;
- time to detect and resolve incidents;
- contractor knowledge transferred to government teams;
- product outcomes for veterans and staff;
- and systems responsibly retired.
These indicators reveal whether investment produces ownership and learning.
The Strategic Inference¶
AI workforce funding is part of the safety and accountability architecture.
Without stable internal teams, agencies cannot maintain evidence, monitor operation, challenge suppliers, or preserve mission knowledge. They become purchasers of outputs rather than owners of public capabilities.
VA’s mission can attract exceptional people. Salary flexibilities and fellowships can open doors. A durable product operating model gives those people a reason and a mechanism to stay.
The governing question is not whether government can hire an AI expert this year. It is whether the institution can sustain the human capacity required to remain accountable for an AI-enabled service throughout its life.
This essay was substantially revised in July 2026 to replace the original budget-news summary with an evidence-based workforce and product-funding analysis. It incorporates VA’s later workforce blueprint.
For related work on AI operating models, product strategy, and institutional capability, see my portfolio or connect on LinkedIn.
References¶
- U.S. Department of Veterans Affairs, “VA Secretary Press Conference,” November 29, 2023.
- DigitalVA, “VA Partnerships Attract Top Industry Talent to Careers in Public Service.”
- Rebecca Heilweil, “VA struggles to bring on AI talent without a long-term budget from Congress,” FedScoop, January 19, 2024. This report was the historical prompt for the original post.
-
U.S. Department of Veterans Affairs, VA Artificial Intelligence Workforce Resources Blueprint, November 2024. ↩