Case Studies of AI in Counseling, Coaching, and Career Matching
The State of play
O’Neil Edwards was running out of options. He knew from counselors, professors, and fellow students at Baruch College that getting an internship was critical to eventually landing a full-time job.
But, as a Jamaican immigrant in one of the most expensive cities in the world, he had to juggle multiple jobs totaling 40 hours weekly while pursuing his degree. His GPA had taken a hit, and he lacked the extracurriculars that fill résumés. Internship applications led nowhere. Two fellowship programs for underrepresented students rejected him.
“I was literally reaching out to anyone who would just hold my hand and just slightly pull me in the right direction,” Edwards says.
Then he found Basta, a nonprofit pairing artificial intelligence tools and human coaches to help first-generation students build professional networks. Through Basta’s program, Edwards connected with Bloomberg and landed a job shortly after graduation. Five years later, he’s a senior account manager.
“Working at Bloomberg has transformed my life,” he says.
The Big Idea
Edwards’ story represents what could be a new frontier in education and workforce development—one in which generative AI and other forms of articial intelligence dramatically expand access to the kinds of counseling, coaching, and career matching that make it possible for students to climb the economic ladder. There are glimmers of that future across the country—from urban centers like Phoenix, where students at one of the nation’s largest universities can get instant feedback on writing, to rural Kentucky, where adults training to be addiction recovery specialists practice client interactions with AI simulations.
And organizations with a national reach, like CodePath and CareerVillage, are particularly keen on the potential of AI to lower the cost of delivering coaching and other support services. “AI is going to be a crucial tool in our scaling,” says Zack Parker, CodePath’s vice president of product engineering.
“The best way to figure out how we respond is to play with AI every day.”
These tools, however, are still in their toddlerhood—as is our understanding of the ways that AI more broadly could impact learning and coaching across higher education and workforce systems. And mixed in with the promising experiments, there have been glimmers, too, of a future that gives pause to researchers and educators who are exploring how AI could impact opportunity and wellbeing.
Julia Freeland Fisher, director of education at the Christensen Institute and author of “Who You Know,” sees promise in well-designed AI tools, but also worries that young people will be tempted to rely on bots and forgo the one-off human interactions that build networks over time.
“The irony is that while those bots could indeed provide useful information, support, and even a sense of connection, they can’t actually open doors to opportunities the way our human weak-tie networks do,” she wrote in a piece for Work Shift. “In short, bots could inadvertently chip away at the very opportunities they’re being engineered to unlock.
That can be avoided, Fisher says, by building AI tools that encourage human interaction and make it easier for people to connect. Our experience with social media, though, shows that building human-centered AI tools would take focused policy and investment, she says. Otherwise, the market tends to incentivize tools that keep people on whatever tech platform for as much time as possible.
In this Work Shift explainer, we highlight several nonprofits who are building AI tools with those concerns in mind. It’s one of several emerging trends in AI and student support that we’ve identified across our reporting. The four are:
- A push among educators, researchers, and nonprofits for AI tools that complement human connections rather than replace them, using technology to facilitate meaningful interactions between students and professionals.
- The development of systems that make implicit professional knowledge explicit, codifying the unwritten rules of career advancement that have traditionally remained hidden from first-generation students and low-income workers.
- A push to leverage AI to move toward skills-first education, hiring, and advancement, identifying transferable competencies that a focus on degrees often overlooks.
- A hope that these innovations extend personalized support to previously underserved populations, making high-quality guidance available on-demand and more affordable to scale.
Nothing about how those trends play out is guaranteed. Within education and workforce, AI is still very much in its “play era”—a time of experimentation with all the hope, uncertainty, false starts, and reboots that implies. The nonprofits and initiatives highlighted in the cast studies below are some to watch, as the field looks for signals of how AI will ultimately impact education, careers, and economic mobility.
Case studies: Counseling, Coaching, Career Matching
Data and Navigation
Matching Hidden Talents to Unexpected Opportunities
AdeptID, a Boston-based public benefit corporation, is pioneering an AI-driven approach to skills-based hiring for workers without four-year degrees. Rather than focusing on credentials or job titles, the company’s algorithms identify transferable skills that traditional hiring methods might miss.
In one striking example, AdeptID’s work with Boston Medical Center revealed that Dunkin’ Donuts cashiers could successfully transition to pharmacy technician roles. The connection isn’t immediately obvious, but both positions require similar competencies: working with point-of-sale software, providing customer service, and handling stressful situations.
“The data showed a high degree of skill overlap and that folks made that transition at a higher success rate than the industry overlap would suggest,” says Fernando Rodriguez-Villa, AdeptID’s CEO and co-founder. “The data is telling a story that the work environment of one is predictive of work in another.”
AdeptID’s play isn’t to create a new layer of hiring tools. Rather, it aims to build a talent-matching infrastructure that can be integrated with any existing tool in the hiring-and-learning space. As a result, AdeptID’s presence is invisible to its end users, in the same way most shoppers don’t know that Stripe works to make sure their credit card transactions are handled securely.
Last year, 11M+ people had used tools that tapped AdeptID by the fall, up from 3.6M people in all of 2023. That means the company has made 11M job recommendations. “The more people use our matching, the more accurate and nuanced we become with our recommendations,” says Rodgriguez-Villa.
The feedback loop helps employers better understand job candidates and fill open roles faster. They also get a larger pool of jobseekers, including skilled workers without four-year college degrees.
The company aims to serve 50% of the U.S. workforce within two years.
Democratizing Career Advice
The arc of career development starts long before the first interview and continues past getting a job. The nonprofit CareerVillage hopes to make that process smoother and more equitable with the help of AI.
The group has developed an AI tool called Coach that provides free career guidance to users at any stage, from K-12 students to mid-career adults looking to upskill or change careers.
One of Coach’s most popular features so far, according to CareerVillage founder and executive director Jared Chung, is its ability to hold live mock interviews—via chat. Help with resume and cover letter writing is also in high demand. However, Coach can also help users break down some less obvious, but still crucial, barriers to employment, including understanding the list of qualifications on a job posting.
What distinguishes Coach from general AI assistants like ChatGPT is what Chung calls its “scaffolding”—a structured approach that helps users learn how to think about careers and develop skills for navigating professional transitions. When users ask questions, Coach responds with clarifying questions about where they are in their career and what specific help they need, guiding them toward more focused inquiries.
The tool also addresses psychological barriers that affect career decisions. For example, research shows women tend to apply for fewer jobs than men, often because they feel underqualified. Coach helps users analyze job qualifications and think through whether they should apply, potentially encouraging more diverse candidates to pursue opportunities they might otherwise avoid.
Developed in partnership with over 20 nonprofits and educational institutions, Coach is being implemented in a range of settings, from school districts to workforce boards to nursing programs. In rural areas like the Berkshires, where expertise on emerging careers may be limited, the tool helps fill knowledge gaps while complementing in-person guidance.
“Students need positive interactions in person, but when that can’t happen, I think online is the next best thing,” says Kat Toomey, a youth program specialist at MassHire Berkshire workforce board.
Learning and Skill Development
Simulating High-Stakes Conversations
In Kentucky, the opioid epidemic is an ongoing public health crisis that not only makes life more precarious for individuals, but hampers economic growth. The same is true in many regions of the country. All told, some 2.7M adults are out of the labor force because of addiction—at a time when there aren’t enough workers to meet demand.
Now, Kentucky is aiming to help people in recovery leverage their experiences to help others confronting addiction. And to turn it into a job—and a source of stability.
The Kentucky Community & Technical College System (KCTCS) and the state’s Healthcare Workforce Collaborative got together to create the Career Ladders in Mental and Behavioral Health (CLIMB-Health), which is about a year in. The program provides a pathway for people in recovery and reentry to quickly earn a certificate that qualifies them for jobs as entry-level peer support specialists serving others in recovery.
A core part of the program is learning how to hold client sessions. That’s where AI comes in. KCTCS uses an AI tool called CONVO, built on the ChatGPT platform, to simulate conversations with clients, allowing students to practice their skills in a safe environment before working with actual people in recovery.
Students specify the type of client they want to practice with—such as a veteran struggling with methamphetamine addiction or an incarcerated person going through divorce—and CONVO generates realistic conversations. After each simulation, the system provides detailed feedback on students’ performance and areas for improvement. Students can share these analyses with instructors or supervisors for additional guidance.
“One of the basic, fundamental pillars of peer support is communication and the other is your recovery story,” says Carl Wilson, senior fellow for healthcare workforce initiatives at Kentucky’s Council on Postsecondary Education.
AI-Enhanced Writing Support
Arizona State University is exploring many different uses for AI, through a major partnership with OpenAI. One especially promising area is writing, a skill that is strengthened through regular feedback and revision.
In a course called Writing for Scholarly Publication, for example, doctoral students used an enterprise version of the ChatGPT platform to receive feedback on their work, creating an experience similar to peer review. Amber Hedquist, a doctoral candidate in the English department, found the tool particularly useful for getting immediate feedback.
“I think I’ll always prefer talking to a colleague or a faculty member who understands my field as a whole to give me suggestions,” she says. “However, the GPT was really helpful in those instances where maybe I couldn’t wait a week for feedback.”
Looking forward, Hedquist sees potential for integrating the platform into undergraduate composition courses, helping students evaluate their work against class rubrics, visualize their writing from different perspectives, and consider alternative word choices. Teaching students how to interact with the technology in a sophisticated way, rather than as a simple “input-output” system, will be critical.
“I think not incorporating AI into the writing classroom is a disservice to students at this point,” she says.
Support & Coaching
Testing Intelligent Tutoring
Artificial intelligence can help instructors with course design and development, giving them more time to work with students. A new project from the American Institutes for Research seeks to bring this promising use of the technology to workforce training.
AIR’s partners for the initiative are Per Scholas, a well-established sectoral training provider, and the University of Memphis. The coalition is working to develop an intelligent tutoring system (ITS) for Per Scholas and its tech training programs. Specifically, instructors will use the adaptive software to help prepare learners for certification exams.
The goal is to reduce the burden on instructors, says Samia Amin, AIR’s managing director for workforce innovation and learning. “It’s so hard to keep up,” she says. “The instructors are chomping at the bit.”
ITS systems have been explored in K-12, but not much in workforce education. Per Scholas is experimenting with the technology in IT training, Amin says, with interest in possibly expanding to the nonprofit’s cybersecurity program.
AIR is taking an ecosystem approach with the technology’s potential in workforce training, with an initial focus on early adopters. “You want the people who will go into the uncertain with you,” says Amin.
The research group will examine results from the project, to see if students fare better on their certification exams and whether it’s cost-effective for Per Scholas. Amin is optimistic that a strong case will emerge for scaling up the use of similar AI tools by sectoral training organizations, whose upskilling services will be in high demand in coming years, she predicts.
“They’re going to be much more important as AI and job augmentation takes place,” says Amin.
Building Social Capital Through Data and AI
For decades, education reform focused on getting more students from disadvantaged backgrounds into college, and then on actually getting them through. More recently though, there’s been a growing recognition that even graduating may not be enough to land an upwardly-mobile job. A key stumbling block is poor social capital—the networks of people who can expose students to a wide range of careers and help connect them to opportunities.
Basta, a New York-based nonprofit, aims to remove that barrier. It’s one of several nonprofits that have come online in the past decade with a specific focus on helping first-generation and underrepresented minority students build social capital. Basta’s particular take leans heavily on data and, increasingly, AI tools.
Its AI-powered Seekr tool, for example, helps students explore their interests and priorities when it comes to work and identifies where they are on their career paths. Crucially, the organization pairs its AI tool with a high-touch approach. Its fellows meet regularly in small groups for career coaching and most importantly to make connections with potential employers in their field of interest. In this way, Basta is using AI to help students more strategically grow and develop their human networks.
Since the organization launched in 2016, it’s worked with around 8K students across more than 60 campuses in the country. With Seekr and a new $4M gift from MacKenzie Scott—about 50% of Basta’s operating expenses for the next year—the organization is planning to grow substantially, particularly in the U.S. South with a focus on minority-serving institutions.
By 2030, they hope to have reached 35K students, with 1M users of Seekr.
Making Interview Prep Available 24-7
CodePath, a nonprofit training provider, exists to help diverse students outside of the Stanfords and MITs develop the technical expertise and career experience they need to land top software engineering roles. The breakthrough on generative AI has affected everything from the skills CodePath needs to teach to its business model.
Coaching is an area the organization zeroed in on early. The support CodePath’s 7K students get with their job searches, the all-important technical interview, and general career mapping is a huge part of what makes them successful. But providing that kind of coaching also is time and staff intensive—making it as expensive as it is essential. An AI-powered career coach can help bring down the cost.
The tech also can be available 24-7 and doesn’t get bored answering the same questions over and over the way a human does. “One of the things that makes it challenging to fill those roles is the conversations can be kind of repetitive,” says Parker, CodePath’s vice president of product engineering. But repetitive tasks are perfect for AI.
To create the coach, Parker and his colleague Katelyn Kasperowicz first built out a series of bots designed to be good with a specific task, such as helping students write a professional email, negotiate a job offer, or prepare for a technical interview. They then created an AI overlay that stitches them all together in a way that appears seamless to the user.
Ultimately, they created a useful tool for students, but also learned a lot as an organization along the way. “A lot of the work we do is not in the code we write, but in the organizational muscle that we build,” Parker says.
AI-Powered Alumni Connections
Protopia started as a tool for fundraisers to engage alumni. But while it’s positioned to tap universities’ sizable fundraising budgets—it’s built around mentoring and career development. The basic premise is simple: A student can ask any question, and Protopia will find a knowledgeable alum to answer it.
By using AI, Protopia can offer participants a relatively light commitment. The company taps into the universe of alumni with an active email address, who don’t have to create an account on a specific platform or sign up for mentoring. A student just asks a question through a simple online form, and the AI model gets to work.
It dissects the question, combs through the alumni database to find the best fits, and then fires off emails on the student’s behalf. If the first-choice alumni don’t answer within a set time period, the model moves on to the next best matches. About 91–93% of all questions get answered—most by 2 to 3 alums—and 10–15% evolve into an ongoing connection or mentorship, the company says.
AI dramatically increases the odds that questions from students yield answers from alums, says Max Leisten, Protopia’s founder and CEO. “If I ask you the right thing at the right time, I know I can get you to help.”
Fisher, of the Clayton Christensen Institute, likes Protopia’s approach to using AI in career support because they are designed to help students connect with other humans, rather than just a bot. While AI can help extend the reach of mentor networks, she says real-life experiences and connections are crucial to reducing unequal access to good careers.
The Kicker: “We could flood that gap with bots and pat ourselves on the back and still have the same inequities,” says Fisher.
EDITOR’S NOTE
This Work Shift Explainer is the second in a series looking at trends in how AI is shaping education and work, and highlighting notable examples of its use. That series is part of our broader AI and Economic Opportunity Reporting Initiative. Our independent reporting for that initiative is supported by Cognizant, GitLab Foundation, Kapor Foundation, and Walmart.
This explainer was written by Elyse Ashburn, Paul Fain, and Colleen Connolly. Erin Strout and Margaret Moffett also contributed reporting.
