EDA webinar outlines data-reporting, evaluation expectations and July 10 deadline for AI Upskill pilot

Justin Tulie, acting director for the Office of Innovation & Entrepreneurship at the Economic Development Administration, told a webinar audience that the agency is using its statutory authority to run an AI Upskill pilot and wants applications that both propose workforce training and enable rigorous evaluation to inform future scaling.

The webinar focused on data and evaluation expectations for the NOFO, which has been posted for over a month. Tulie said EDA will use application materials and post-award reporting — and will contract an external evaluator — to understand how sectoral partnerships and training affect both business and worker outcomes.

EDA staff emphasized that every application must identify a regional "special need" for AI training, documenting (1) the economic significance of the industry in the region, (2) current employer use of AI with concrete examples, (3) why AI is important for competitiveness, and (4) the specific workforce skills gap and how many workers will need upskilling. "All proposals must identify a special need," said Carla Garrett, who runs EDA's workforce team, explaining that applicants should show how activities directly address that need.

Applicants must propose sectoral partnerships that include employers who will validate occupational profiles, participate on advisory groups or pilot training, and provide commitments that the NOFO will use to assess feasibility. Garrett advised including a table of employer partners with key details (name, address, EIN) and personalized commitment letters on company letterhead; she said stock language is insufficient to show credible commitment.

Tulie warned applicants to avoid overly optimistic outcome promises that lack supporting commitments or realistic implementation plans. "We are going to hold awardees accountable for what they put in their application," he said, noting that application targets are incorporated into post-award scopes of work and monitored for up to five years after award start.

EDA described the data-collection approach it expects from grantees: an annual workforce-collection tool run in partnership with the Census, an employer questionnaire to assess business outcomes, and pre/post participant questionnaires to measure training impacts on familiarity, skills and wages. Awarded projects may operationalize reporting through state systems or partners; EDA said grantees may use award funds to support data reporting and that program officers will provide technical assistance.

On evaluation, Tulie said the NOFO favors projects that can be evaluated rigorously during and after the period of performance to support replication and scaling. He added that EDA seeks openness to evaluation rather than fully developed research protocols at the application stage: "We want to see your openness to evaluation," he said.

Presenters addressed common applicant questions: EDA is agnostic about employer size (large, medium or small) but seeks a diverse award portfolio across industries; protections exist for individual-level PII through the Census workforce collection; procurement for contracts must follow public procurement rules; legal clarifications (for example on whether curricula become public domain) will be posted to FAQs or handled via the OE inbox.

The webinar closed by directing applicants to recorded webinars, slide decks and FAQs on EDA's website and reminding attendees to begin organizational registration early. The application deadline for submissions through EDA's platform is July 10, 2026, at 4:59 p.m. Eastern. The webinar concluded with presenters saying the agency will support applicants with technical assistance and updated FAQs.