Bridging the Readiness Gap

The Workforce Development system cannot meet the demands of its stakeholders without a fundamental shift in how it delivers and measures career readiness.

Workforce Development in the United States faces an important crossroads, as the workforce confronts several major transitions at once. Between the current state of the labor market, the rise of Generative AI and increased technological acceleration, and questions about the return on investment of the traditional four-year college degree, workforce development faces one of the most challenging and consequential inflection points in recent history.

The Workforce Development infrastructure is considered the lynchpin that ensures that job seekers across the spectrum are prepared to enter (or re-enter) the workforce. This population of job seekers is incredibly large and diverse, a multi-generational array spanning from returning citizens and transitioning military service members to recent college graduates entering a softened labor market with entry level requirements that are remarkably different from even two years ago. However, it has faced cascading challenges over the past 20 years, where severe funding cuts have led to case load swelling, increased administrative burden, and subsequent burnout within their organizations.

Corporate employers have expressed dissatisfaction with entry-level employee readiness, saying they lack both the technical and soft skills to meaningfully contribute to their companies. This has translated into increased turnover rates and higher workloads for their existing employees. Substantial data suggests that focused practice on both technical and soft skills increases graduate readiness, but adoption has been slow due to a combination of case worker overwhelm and the lack of solutions that focus on this use case.

All these problems are interconnected, so the solution will not be a single tool or piece of technology. Rather, the solution will have to be systemic, one that involves scalability, improved synergy between employer needs and program curricula, strong and focused practice infrastructure, and quantifiable, provable outcomes for its graduates. These elements work together because scalability addresses the capacity gap, improved employer alignment closes the readiness gap, the focused practice equips graduates with the necessary skillset to contribute to their new role, and quantifiable outcomes satisfy the accountability standards with clear pathways for improvement.

While the challenges for the workforce development infrastructure are daunting, there is also a massive opportunity to meet this moment and create transformative solutions for job readiness across the United States.

As of March 2026, the labor market faces an unprecedented mixture of factors that has affected job seekers across the United States. The first factor is the current state of the labor market itself. According to the Bureau of Labor Statistics, the job market has been largely frozen in place for the past 16 months. I will use the period between 2015 and 2019 as the baseline for a "normal" job market in the United States, since that reflects the point between the post-recession period and the beginning of the COVID pandemic in 2020.

Although the overall layoff rate has remained at an all-time low of 1.1%, the hiring rate has remained steady at 3.3%, which has created a level of stasis within the labor market. In fact, aside from the post-COVID hiring boom in 2021 and 2022, overall hiring rates have regressed to the same levels as 2010, when the United States was recovering from the effects of the Great Recession.

Another set of interesting figures here are what are called the quit rate and the ghost job rate. The quit rate, which is defined as the frequency of employees voluntarily leaving their jobs, is used as a relative barometer of labor market health. When the quit rate is high, this means that employees feel confident that they can find better fitting roles with a higher salary elsewhere. When the quit rate is low, it means that employees feel insecure about the broader labor market, which makes them more likely to stay in their current role. During the post-COVID hiring boom, the quit rate peaked at 2.7%, which indicates that a relatively large share of the workforce felt confident enough in their prospects to voluntarily leave their current role to pursue another. Since that 2022 peak, the quit rate has regressed to 2.0%, which is below the pre-pandemic average of 2.1% to 2.3%.

A growing number of job opportunities that are posted never actually get filled with a new employee, which has birthed the tracking of what are called "ghost jobs." Starting in 2019, a workforce intelligence firm called Reveilo Labs began tracking ghost job rates, which is measured by how many hires are made per ten job postings. For example, 8 out of every 10 open, advertised roles were filled with new employees. That means the ghost job rate was 20% in 2019. By the end of 2025, that ghost job rate has tripled to 60%, meaning that over half of open, advertised roles never actually get filled with new employees.

This has major implications for the actual state of the labor market. The official job opening rate at the end of 2025 was 4.2%, which is well below the 2022 post-COVID peak of 6.8%, but still within the range of the pre-pandemic baseline. If the ghost job rate is 60%, that means the "real" job opening rate is 1.68%, which is less than half of the pre-pandemic baseline and even lower than the 2.2% figure posted during the Great Recession in 2010.

The share of long-term unemployment is another factor that can be used to measure overall labor market health. In this case, the long-term unemployment rate is defined as the share of unemployed workers that have faced continuous unemployment for at least 27 weeks. Using the pre-pandemic baseline of 2015-2019, long-term unemployment hovers from 21% to 25% of the total unemployed labor force. Like other major labor factors, the long-term unemployment rate dramatically spiked during COVID, peaking at 40% in 2021 before falling to 20% in 2023. Since then, the long-term unemployment rate has steadily risen to 25.3% as of February 2026, with no signs of an immediate dip.

The average (mean) duration of unemployment has followed a similar pattern, with the pre-pandemic baseline settling at 22 weeks in 2019 and peaking at 30 weeks in 2021, before returning to 21 weeks in 2022. Since then, the average (mean) duration of unemployment has steadily risen to 25 weeks as of February 2026, its highest rate since 2014.

What these figures describe, in practical terms, is a labor market that has stalled. Workers who are employed largely feel unable to leave, and workers who are unemployed are finding that the path back to employment is longer and more uncertain than it has been at any point in the past decade outside of the COVID period. The indicators that economists use to measure a healthy labor market, including hiring activity, quit rates, and the ratio of genuine job openings to unemployed workers, are all pointing in the same direction. Conditions have been softening steadily since 2022, and as of early 2026, there is no data to suggest that trend will reverse soon.

According to the World Economic Forum's 2025 Future of Jobs Report, 86% of surveyed employers see Artificial Intelligence (AI) as likely to drive business transformation by the year 2030. This development, which started with the release of ChatGPT in 2022, has already changed the landscape of the workforce and will have even more significant effects going forward. The rise in demand for generative AI skills can be seen in the spike of enrollment for courses that focus on building those skills. According to a Coursera analysis, consumer enrollment for generative AI courses rose from 10,000 in April 2023 to 240,000 in October 2024, a stunning 24x increase in 19 months. Enterprise enrollment for generative AI courses saw a similar increase, rising from 2,000 enrollments in April 2023 to 180,000 enrollments in October 2024.

This investment in skill development has come with the expectation that workforces that adopt generative AI increase their productivity and that individuals will attract more lucrative career opportunities. Unlike other technical innovations that have fundamentally changed the nature of blue collar, trade-driven work, generative AI has its most transformative changes on white collar, office-driven work.

Almost all occupations have some exposure to generative AI capabilities, but the industries that project to have the greatest generative AI exposure are those that heavily rely on data processing and information processing. According to a study conducted by Upjohn Institute for Employment Research, roles that normally do not require college degrees, such as office clerks, customer service representatives, and data entry keyers, are most prone to full automation and major decline for job demand. This is due to these roles having "characteristics such as routine cognitive tasks, structured environments, and limited demand for advanced judgment that align closely with areas where generative AI and related information technologies have made significant inroads."

There are two important distinctions this research covers. First, occupations with high risk to disruption from AI aren't limited to those that are considered "low wage" or "low skill"; the tables show a litany of "high skill" occupations, such as airline pilots and mechanical engineers, with high exposure to disruption. Second, demand for many, if not most, of these high exposure occupations are still projected to grow over the next ten years, contradicting the claims that high skill occupations are disappearing due to AI. The more likely consequence of this disruption will be significant adjustments to the expectations of these occupations and the output quantity and quality of a prospective job seeker. These adjustments will require structured reskilling and upskilling, which will increase pressure on employees looking to advance and stand out within their occupations.

A study conducted by the St. Louis Federal Reserve Bank shows that "among workers who used generative AI in the past week (21.8% of all workers), between 6.0% and 24.9% of all work hours were assisted by generative AI. This means that, on average, generative AI is not just an occasional tool for its users but also an integral part of their work routines."

This study also shows that workers that are primarily in management and math/computer centric fields tend to use generative AI for the highest share of their tasks.

This embrace of generative AI in the workforce has already had a material impact on entry level roles in occupations exposed to AI disruption. An expansive research paper called "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence" from Stanford's Digital Economy Lab examined the effects of generative AI on the current labor market. Their study showed that occupations with high generative AI exposure have experienced outsized declines in hiring job seekers aged 22-25 compared to other age groups, with the study showcasing software engineering and customer service as high profile cases. Their study also showed that the sharpest hiring decline occurred in occupations where AI could automate the bulk of the tasks, as opposed to augmenting task completion.

This disruption in entry level roles has adversely affected college graduates who have entered a labor market that is radically different from what it was when they entered college. The effects of generative AI have had the compounding implications of decreasing entry level roles for college graduates and increasing the skill and experience thresholds for a competitive entry level candidate. High school graduates are taking notice of this shift, and it is challenging their perceptions of the value of a four-year college education.

The consequences of the increase in AI adoption in the workplace have materially influenced the decision making of current college students and high school upperclassmen, particularly those who would normally be competitive for placement in four-year colleges and universities.

According to the Lumina Foundation-Gallup State of Higher Education, 42% of bachelor's degree students and 56% of associate degree students have given at least a fair amount of thought to change their college major due to concerns about AI. In that same study, a full 16% of college students have changed their college major and attributed their reason to the impact AI may have on their industry.

Anecdotally, my position in the Career and Technical Education (CTE) committee allows me to view a hyperlocal shift in high school seniors who are opting to attend two-year colleges or going straight into the workforce instead of attending universities. The perspective that I have observed locally has played out nationally, with a survey of 5,000 students conducted by EAB revealing that "43 percent of students say that AI will influence the career they pursue" and "39 percent say that AI is pushing them to consider alternatives to college, including starting a business or entering an apprenticeship." This shift has raised concerns about my school district's ability to support these students with the same robustness they would if their goal was to attend a four-year university, and it presents a reasonable assumption that school districts throughout the United States share a similar struggle.

The combination of these factors has driven a sea change in the perception of the value of college in general, accelerating a 15 year trend where the percentage of Americans who do not think college is important has risen from 1% in 2010 to 30% in 2025.

These forces are converging on a single problem. A stalled labor market, the disruption of entry-level white-collar work by generative AI, and a generation of students questioning the value of a four-year degree have left a growing share of Americans without a clear path to economic mobility. The pathways that once delivered reliable outcomes are no longer doing so at scale, and the systems built to move workers into the economy have not adapted at the same pace. That gap is what workforce development now exists to close, and the demands on it are growing faster than the field has been resourced to meet.

On July 4, 2025, Congress passed and the President signed a budget reconciliation bill that included the largest expansion of federal financial aid for short-term job training in a generation. The provision, known as Workforce Pell, extends Pell Grant eligibility to career training programs that run between 150 and 599 clock hours over 8 to 15 weeks. The Department of Education published its final rule on May 19, 2026, and the program took effect on July 1, 2026. For the first time, students pursuing short-term credentials in fields like healthcare, information technology, and the skilled trades can access the same federal grant aid that has supported traditional college students for more than fifty years.

The funding expansion is only half the story. The other half is accountability, because Workforce Pell is the most outcome-driven federal aid program ever created. To become eligible and stay eligible, a program must clear three thresholds. It must maintain a completion rate of at least 70%, measured within 150% of the program's normal length. It must maintain a verified job placement rate of at least 70%, measured in the second quarter after completion. And it must pass a value-added earnings test, meaning the median earnings of its graduates must exceed the program's tuition and fees plus 150% of the federal poverty level. These are not self-reported numbers. Placement and earnings must be verified through state administrative data, including unemployment insurance wage records.

The penalty for failure is severe. A program that falls below the completion or placement thresholds loses eligibility for two years, and the lockout extends to "substantially similar" programs that share the same instructional and occupational classification codes. Approval itself runs through two gates. The state's governor, in consultation with the state workforce board, must certify that a program prepares students for high-skill, high-wage, or in-demand occupations and meets the hiring requirements of employers. The Secretary of Education must then separately verify the program's outcomes.

I refer to this as the accountability moment because it crystallizes a pressure that has been building across the entire workforce ecosystem for years. The 70/70 thresholds convert career readiness from a mission statement into a measurable, auditable, and fundable standard. Providers that can prove seven out of ten students finish and seven out of ten get hired will unlock a durable new funding stream. Providers that cannot prove it will be locked out. Every workforce organization, community college, and training provider in the United States now has a direct financial incentive to do two things exceptionally well: prepare graduates who actually get hired, and produce the data that proves it. The rest of this paper examines whether the system is currently equipped to do either.

The workforce development system is the connective tissue between the labor market described in Part One and the millions of Americans trying to enter it. The modern system operates under the Workforce Innovation and Opportunity Act (WIOA), which Congress passed in 2014. WIOA funds six core programs serving adults, dislocated workers, and youth, delivered through a network of state and local workforce development boards and roughly 2,400 American Job Centers across the country. The Adult, Dislocated Worker, and Youth programs alone serve hundreds of thousands of participants each year, and the broader system, including Wagner-Peyser employment services, serves millions more. The population it carries is exactly as diverse as this paper's opening described: transitioning service members, returning citizens, dislocated mid-career workers, opportunity youth, English language learners, and now a rising wave of college graduates and degree skeptics who cannot find a foothold in a frozen labor market.

The system carries this mandate with a fraction of its historical resources. Funding for WIOA's three primary formula grant programs totaled about $3.3 billion in fiscal year 2023, which according to Jobs for the Future is down about 50% from fiscal year 2000 after adjusting for inflation. The longer arc is even starker. Federal investment in workforce development peaked in the late 1970s, when spending on the Comprehensive Employment and Training Act reached over 7 billion in 1979, the equivalent of roughly $60 billion today. Harvard economist Harry Holzer, testifying before Congress, calculated that federal spending on workforce services and training has dropped by two-thirds since that peak, even as the labor force grew by about 50%. Jobs for the Future estimates that matching the late-1970s investment would require raising annual WIOA formula allotments to about $35 billion, and that the United States would need to spend an additional $80.4 billion per year on employment and training just to match the average of other OECD countries as a share of GDP.

The practical consequence is that very little money reaches actual training. Of the roughly $4 billion in WIOA Titles I and II, only about $0.5 billion is spent on training, reaching about 220,000 people per year at just over $2,200 per trainee. Compare that to Pell Grants, which provide up to nearly $7,000 per year to 6 million students in higher education. The workforce system is asked to prepare the hardest-to-serve populations in the country with less than one-third of the per-person investment that traditional college students receive.

Chronic underfunding is compounded by chronic instability. In Fall 2025, the National Association of Workforce Boards surveyed 139 workforce boards across 33 states and the U.S. Virgin Islands and found that 64% had cut costs due to funding uncertainty, through actions that included staff layoffs, site closures, reduced programs, hiring freezes, and delayed payments. Board leaders reported that they cannot sign leases, hire staff, or commit to program contracts without predictable funding. The same survey found demand moving in the opposite direction. Boards reported serving more individuals experiencing unemployment and food insecurity, and 55% reported changes in customer interaction driven by new SNAP work requirements, which push more people into a system that is actively shrinking.

Inside the organizations, the math lands on individual case workers. Caseloads have swelled as funding has fallen, and the participants on those caseloads increasingly arrive with complex, overlapping needs spanning housing, transportation, childcare, and digital access. The same capacity crisis exists on college campuses, where NACE benchmarking data puts the ratio at one career services professional for every 2,263 students. Gallup has found that only 43% of students who used their career services office found it helpful, and only about half of graduates report visiting at all. When one human being is responsible for guiding thousands, guidance becomes triage.

Administrative burden absorbs much of what capacity remains. WIOA's reporting requirements are extensive, and its performance measures are calculated only after a participant exits a program, a structure that Jobs for the Future notes actually discourages continued service once a participant lands any job, regardless of quality. Case workers spend hours documenting services instead of delivering them. The result is the cascade described at the opening of this paper: swollen caseloads, rising administrative load, and burnout across the organizations the system depends on. The system is being asked to close the widest readiness gap in decades with half the funding it had twenty years ago. To understand exactly what it is being asked to close, we have to look at what employers say they need.

Employers and educators are describing two different graduates. The Cengage Group 2025 Graduate Employability Report found that 89% of educators believe their students are prepared to enter the workforce, while almost half of graduates themselves, 48%, say they feel unprepared to even apply for entry-level jobs in their field. The two groups do not even agree on what preparation means. Employers ranked job-specific technical abilities as their top hiring priority. Educators ranked those same skills last, prioritizing soft skills like critical thinking instead. Graduates are caught in the middle of that disagreement, and the outcomes show it: only 30% of 2025 graduates found entry-level jobs in their fields, and 76% of employers reported hiring the same or fewer entry-level employees than the year before.

Employer sentiment data tells the same story from the hiring side of the desk. General Assembly surveyed 651 company leaders and 2,361 employed adults across 2024 and 2025. Only 22% of leaders said entry-level employees were very or completely prepared to do their jobs, and among US leaders that figure fell to 18%. Nearly a third, 31%, described entry-level workers as hardly or not at all prepared, and 29% of leaders said they would avoid hiring entry-level employees altogether. Perceptions are not improving with time, either: 53% of leaders said today's entry-level employees are less prepared than those of five years ago.

When asked why, employers overwhelmingly point to soft skills. In the General Assembly research, 56% of leaders blamed weak soft skills for entry-level unpreparedness, up from 50% the year before, and at companies with more than 1,000 employees that figure reached 64%. NACE research quantifies the perception chasm with precision. Employers and graduating students agree that communication, critical thinking, and teamwork are the most important career readiness competencies. But 84.6% of students rate themselves as very or extremely proficient in professionalism, while just 50% of employers agree. Graduates are not lying about their readiness. They have simply never received calibrated feedback against a real employer standard, so they have no way to know where they actually stand.

The most troubling finding is that this gap has no owner. In the 2025 General Assembly survey, 78% of leaders said employees themselves are responsible for arriving with the skills to succeed, up from 74% the year before, while fewer pointed to employers or educational institutions than in the prior year. While the parties debate responsibility, the costs land exactly where this paper's introduction placed them: higher turnover, heavier workloads for incumbent employees, and hiring managers who respond by raising experience thresholds, which shuts out the very candidates the workforce system serves. This is the gap workforce development is now expected to close, with the resources described in the previous section. The good news is that the field already knows what closes it.

The uncomfortable truth about the readiness gap is that we already know how to close it. The most rigorous evidence in the field comes from sector-focused programs that combine technical training, intensive soft skills practice, and direct employer engagement. Year Up, evaluated through the federally sponsored PACE randomized controlled trial, produced a 30% increase in average annual earnings, more than $8,000 per year, that persisted through seven years of follow-up, with participants accumulating over $38,000 in additional earnings compared to the control group. That is the largest sustained earnings impact ever recorded for a US workforce program in a randomized trial. Per Scholas, which pairs 15 weeks of IT training with career readiness coaching and placement services, produced cumulative earnings gains of roughly 16%, about $42,000 per participant, over ten years of follow-up.

The evidence on focused practice specifically is just as strong. A randomized controlled trial published in the Journal of Political Economy in 2023 found that structured workplace soft skills training produced 13.5% productivity gains and a 256% net return to the firm within eight months of program completion. J-PAL's review of the experimental literature reaches the same conclusion: soft skills training can improve employment outcomes. Research on virtual mock interviews finds that students rate the practice as useful for performing better in real interviews, and that the level of preparation before the simulation is the primary factor driving positive outcomes. The common thread across all of it is repetition. Graduates get better at interviewing by interviewing, and better at workplace communication by communicating, with feedback, many times, before the stakes are real.

The roadblock is arithmetic. Year Up costs about $28,200 per participant. Per Scholas costs about $5,800. The public workforce system spends just over $2,200 per trainee. The proven models cost between roughly three and thirteen times what the system can actually spend per person, because they depend on one-to-one human intensity, and human intensity is precisely what the system has lost. Programs built on manual coaching cannot scale to a system serving hundreds of thousands of people with flat budgets and case workers already at capacity. This is the real technology gap in workforce development: the field's most effective interventions are its most labor-intensive, and labor is its scarcest resource.

The second roadblock is data infrastructure. Workforce Pell requires placement rates verified against state wage records, yet as the Data Quality Campaign points out, most institutions have no direct access to that administrative data and no established pipeline to obtain it. WIOA's exit-based measures were built for compliance reporting, not continuous improvement. Many providers still run intake, coaching notes, and outcomes tracking across disconnected spreadsheets. An accountability regime is arriving faster than the data systems needed to satisfy it.

Technology adoption is accelerating, but unevenly. NACE's 2026 benchmarking poll found that 86% of college career centers now use AI as an assistive tool when working with students, up from just 20% in 2023. Jobs for the Future has documented tech-enabled, data-empowered workforce boards using modern tools to expand reach and improve decisions, but those boards remain the exception rather than the rule. The tools now exist to give every job seeker unlimited structured practice and to give every program auditable outcomes data at a cost the system can actually afford. The organizations that adopt them will be the ones able to meet the 70/70 standard. That conviction shapes the recommendations that follow.

No single tool solves a systemic problem. Based on the evidence assembled in this paper and on our own work deploying readiness infrastructure with workforce organizations, Capstone Tech recommends that leaders evaluate every program investment against four connected requirements: scalability, employer alignment, focused practice, and provable outcomes. Each one addresses a specific failure point documented in the preceding sections, and they only work together. Scalability without alignment produces volume with no relevance. Alignment without practice produces awareness with no skill. Practice without measurement produces effort with no proof. And under Workforce Pell, proof is now the price of funding.

Scale capacity without scaling headcount. Any solution that adds administrative burden to a case worker's day will fail, no matter how good its features are, because staff time is the system's binding constraint. Technology should multiply staff rather than tax them, allowing one case worker to supervise practice and preparation at volumes no human could deliver manually. This is achievable today. In a nine-week deployment with NPower, a national workforce development nonprofit, our Capstone Workforce platform delivered 245 structured mock interviews with zero added staff, work that would have cost the organization up to $24,500 in staff labor to deliver by hand. The benchmark for any workforce technology should be that simple: does it raise the number of participants each staff member can meaningfully serve?

Build programs backward from employer hiring requirements. The Cengage data shows employers and educators ranking skill priorities in nearly opposite order, and that misalignment is a design failure, not a talent failure. Workforce Pell now requires governors to verify that eligible programs meet the hiring requirements of employers. Leaders should treat that as a design principle rather than a compliance checkbox. Practice scenarios, feedback rubrics, and curricula should map directly to the competencies employers actually screen for, and employer partners should review those rubrics on a recurring basis. When the practice standard is the hiring standard, graduates stop being surprised by interviews.

Make focused practice the program, not an add-on. The randomized evidence is unambiguous: deliberate, repeated practice with feedback is what moves both readiness and productivity. Every participant should complete multiple full interview repetitions and workplace communication repetitions, with structured feedback after each one, before their first real interview. The NACE data showing graduates rating themselves far above employer assessments points to the corrective: calibrated feedback against an employer standard, delivered early and often enough for participants to close the gap while it is still practice.

Instrument everything, and prove it. Programs should capture skill growth session by session, alongside completion and placement, in formats that satisfy WIOA reporting, Workforce Pell's 70/70 thresholds, and funder due diligence at the same time. Outcomes data should serve two purposes at once: a compliance artifact for the programs that fund the work, and an improvement loop for the staff who deliver it. The two-year Workforce Pell lockout makes outcome blindness an existential risk, but the opportunity is bigger than the risk. A program that can show a funder verified completion rates, placement rates, and skill growth curves will out-compete every program that shows a testimonial.

Finally, every recommendation above must pass a cost test. Solutions have to work at the system's real economics, which run closer to $2,200 per trainee than to the $28,000 boutique models that produced the field's best evidence. The mandate for technology in workforce development is precisely this: deliver the mechanisms behind the proven models, structured practice, employer-aligned feedback, and verified outcomes, at a per-participant cost the public system can actually sustain.

The workforce development system remains the lynchpin that determines whether job seekers across every generation can enter or re-enter the American economy. That has not changed. What has changed is everything around it. The labor market has frozen, generative AI has reset the expectations placed on entry-level workers, a generation of students is questioning the four-year degree, and the system is being asked to absorb all of it with half the funding it had twenty years ago. This is the most fluid and consequential moment the field has faced in recent history, and the pressure to meet it is no longer abstract.

Workforce Pell makes that pressure concrete. For the first time, accountability and opportunity have arrived together. The 70/70 thresholds reward exactly what job seekers have always needed programs to deliver: graduates who finish, get hired, and earn more than the training cost them. Organizations that build scalable, employer-aligned, practice-driven, and provable programs will not just survive this standard. They will grow because of it, funded by the largest expansion of short-term training aid in a generation.

The opportunity is generational in both senses of the word. A returning citizen rebuilding a career, a transitioning service member translating military experience, and a new graduate facing a frozen entry-level market all need the same thing: focused preparation, honest feedback, and proof of their readiness that an employer will trust. I built Capstone Tech because I believe that proof can be delivered at scale, and the evidence in this paper says the same. The readiness gap is wide, but it is not a mystery. We know what closes it. The organizations that combine scalability, employer alignment, focused practice, and quantifiable outcomes will define the next decade of workforce development in the United States, and the people they serve will be the reason it was worth building.

Shawn Gregoire
Founder and CEO, Capstone Tech

Adoption and scale

• Bick, Blandin & Deming, "The Rapid Adoption of Generative AI" (NBER Working Paper 2024-027C, St. Louis Fed). stlouisfed.org

Occupational exposure

• Eloundou, Manning, Mishkin & Rock, "GPTs are GPTs: Labor Market Impact Potential of LLMs" (Science, 2024). science.org. Preprint: arxiv.org. OpenAI summary: openai.com.
• Brookings Institution, "The Geography of Generative AI's Workforce Impacts." brookings.edu
• BLS / Upjohn Institute, "AI Exposure and the Future of Work." research.upjohn.org

Writing and professional tasks

• Noy & Zhang, "Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence" (Science, 2023). science.org. Preprint (SSRN): papers.ssrn.com. MIT News summary: news.mit.edu.

Knowledge work and consulting

• Dell'Acqua, McFowland, Mollick, Lakhani et al., "Navigating the Jagged Technological Frontier" (HBS Working Paper 24-013). hbs.edu. Full PDF: hbs.edu (PDF). SSRN: papers.ssrn.com.

Customer service

• Brynjolfsson, Li & Raymond, "Generative AI at Work" (Quarterly Journal of Economics, 2025). academic.oup.com. NBER (free): nber.org. Preprint: arxiv.org.

Software development

• Peng, Kalliamvakou, Cihon & Demirer, "The Impact of AI on Developer Productivity: Evidence from GitHub Copilot" (arXiv, 2023). arxiv.org

Aggregate productivity

• St. Louis Fed, "The Impact of Generative AI on Work Productivity." stlouisfed.org
• St. Louis Fed, "Generative AI, Productivity and the Future of Work." stlouisfed.org
• Penn Wharton Budget Model, "The Projected Impact of Generative AI on Future Productivity Growth." budgetmodel.wharton.upenn.edu

Employment effects

• Georgia State University, "Generative AI Boosts Job Growth and Productivity." news.gsu.edu

Human-AI collaboration limits

• Microsoft, New Future of Work Report 2025. microsoft.com (PDF)

Inequality and workforce development

• Social Finance Institute, "Shaping the Future of Work: Generative AI, Inequality, and Opportunity." socialfinance.org
• World Economic Forum, Future of Jobs Report 2025.