Applied AI for Federally Funded Programs
Spence Consultants installs the responsible use policy and the staff training that let WIOA, SCSEP, and other federally funded workforce programs adopt AI without creating a finding at your next review.
Thirty minutes, no deck. We start with what your staff are already doing, because that is the part that carries the risk.
Where most programs are right now
Almost every workforce program we talk to is sitting in one of these. None of them is a governed position, and the difference between them is not how much AI is being used. It is how much of it you can see.
Leadership told staff not to use AI tools. Nothing was written down, nobody was trained, and no one checked. Use moved to personal phones and personal accounts.
The exposure Participant information leaving your systems entirely, on devices you do not control and cannot produce records from.Staff use AI to draft case notes, employer letters, and outreach. It is genuinely helping. There is no approved tool list, no review step, and no record of who checked what before it entered a file.
The exposure No way to show a reviewer that a human made the decision, or that a case note reflects what actually happened.DOL has told the workforce system to build AI literacy, in TEGL 03-25 and the AI Literacy Framework issued in February 2026. You are being pushed to put AI in front of participants. Nobody has told you how staff should handle participant data while doing it.
The exposure Adoption moving faster than governance, with the gap sitting entirely on the grantee rather than the funder.The standard we build to
Not a compliance checklist we invented. These are the questions that follow naturally from the rules you already operate under: participant privacy, human decision-making, documented training, and nondiscrimination. If your program can answer all five with a document, you are governed.
Which participant information has a staff member entered into an outside AI tool?
An approved tool list, paired with a written boundary that names exactly what participant data may never leave your systems.
Who reviewed this AI-drafted case note before it entered the participant file?
A human-in-the-loop rule written task by task, with the review step recorded where the record is created rather than reconstructed later.
Where is your written AI policy, and when was staff trained on it?
A signed, dated policy plus a training roster with attendance, so the answer is a retrieval task instead of a memory exercise.
How do you know an AI tool is not affecting who gets served or referred?
A prohibited use list that puts eligibility, selection, ranking, and referral decisions outside the tool, in writing, before anyone is tempted.
What happens when someone uses a tool that is not on the list?
A named policy owner, an exception request path, and a corrective action step, so the policy is an operating control and not a statement of intent.
The engagement
Scoped around deliverables, not billable hours. Seven weeks from kickoff to signed acknowledgment forms, plus a check-in thirty days later. You end with documents you own, staff who have been trained on them, and a named person inside your organization who runs it after we go.
Phase one Weeks 1–2
Before anything gets written. Most programs are surprised by this part, because the honest inventory rarely matches what leadership believes.
Phase two Weeks 3–5
Specific to your funding stream and your workflows. A generic AI policy pulled from a template fails the moment someone asks how it applies to an eligibility determination.
Phase three Weeks 6–7
A policy nobody was trained to operate is not a deliverable. Staff learn what they can do, supervisors learn what they have to check, and both sessions produce a record.
Investment and timeline
You should not have to book a call to find out whether this fits your budget. Every engagement is a fixed fee against a defined deliverable list, invoiced half at signing and half at delivery. No hourly billing, no change orders for questions.
Start here
$4,800Fixed fee
Two weeks. Phase one only.
Most programs
$14,500Fixed fee
Seven weeks. All three phases, one funding stream.
Multi-program
From $26,000Scoped fixed fee
Ten to twelve weeks. Multiple streams or sites.
At $4,800 it sits below the federal micro-purchase threshold, which is $15,000 in the FAR for awards made on or after October 1, 2025. Under 2 CFR 200.320 a recipient or subrecipient may also self-certify a threshold up to $50,000 annually. For most grantees that means this can be procured without competitive quotes. Confirm against your own procurement policy, since entities may set a lower threshold than the federal maximum.
WIOA Title I local areas may spend up to 10 percent of their allocation on administration. SCSEP grantees are generally capped at 13.5 percent. This is a staff development and compliance cost, and identifying the right line is part of phase one.
An annual review is available at $3,600. Policy refresh against new funder guidance, an updated tool list, and a refresher session for new hires, so the documentation in your file is never a year out of date.
Who this is built for
The work is the same discipline applied across different funding streams. What changes is which rules govern the participant data and who reviews you.
Why us
Lamont Spence was Director of the Urban Seniors Job Program at the National Urban League, a SCSEP national grantee, responsible for seven affiliates across six states and $15 million in grant funds. The program exceeded its negotiated placement and service level goals and came through DOL federal monitoring with zero findings.
Precedent
He wrote the AI case note policy for that program. He also wrote the standards governing AI-assisted resume writing and interview preparation for participants. Both were authored inside a federally funded workforce program, under the rules you operate under.
That is the difference on this page. The second question in the ledger above, the one about who reviewed an AI-drafted case note before it entered the participant file, is not a hypothetical we constructed for a sales argument. It is a question he has already had to answer in writing, for a national grantee, with a monitoring review ahead of him.
It also means the training is calibrated to the staff you actually have. Case managers with full caseloads, supervisors who were promoted for field competence, and front-line workers across a range of digital comfort levels. Nobody leaves the room being told to become an AI engineer.
Spence Consultants is based in Mount Vernon, NY and works with organizations across New York State and beyond.
Before you book
A ban you cannot evidence is not a control, and in practice it is the worst of the three positions. Staff who need the help move to personal accounts on personal devices, which is exactly where you have no visibility, no records, and no way to answer a question about participant data. A governed yes protects the program better than an unenforced no.
Your funder has already spoken, just not in the direction people expect. TEGL 03-25 encourages local boards to use WIOA Title I funds for AI literacy, and DOL issued its AI Literacy Framework as TEN 07-25 in February 2026. Both push AI toward your participants. Neither tells you how your staff should handle participant data in the process. That asymmetry is the whole problem: adoption is being encouraged on a federal timeline, and governance is being left to you. A policy written ahead of a review is a short, fixed-fee project. One written during a review is a scramble, and it cannot retroactively cover the use that preceded it.
This is not a software purchase. There is no license to procure, no system to migrate, and no vendor contract at the end of it. The deliverables are a policy, a tool list, and two training sessions, at a published fixed fee. The Readiness Assessment is $4,800, which sits under the federal micro-purchase threshold. That threshold is $15,000 in the FAR for awards made on or after October 1, 2025, and under 2 CFR 200.320 your organization may self-certify a threshold up to $50,000 annually. WIOA Title I local areas may spend up to 10 percent of their allocation on administration, and SCSEP grantees are generally capped at 13.5 percent. Finding the right line is one of the first things we look at together.
Book a free consultation
Tell us what your staff are already doing and which funding stream you report under. We will tell you plainly whether this is a real gap for you, including when the answer is that you are further along than you thought.