Clara Shih built an AI agent that a million small businesses used to answer customers, remember their preferences and close sales without hiring anyone new. Business owners loved it. Then they told her they no longer needed the staff who used to do that work.
That single piece of feedback helped push her to quit her job this spring. Shih, 44, had spent two decades building AI at Salesforce and Meta. In April she launched a nonprofit betting she can fix the youth unemployment crisis her own industry helped accelerate. By the federal government’s own count, 5.6% of recent college graduates are out of work and 41.5% are stuck in jobs that never required a degree.
The Résumé Behind the Foundation
Shih is not new to artificial intelligence. She cofounded Hearsay Systems, a client engagement platform for financial advisers and insurance agents, and made the Forbes 30 Under 30 list in 2012 as its chief executive. Yext, the digital platform company, acquired Hearsay in 2024.
By then Shih had already moved on twice. She became CEO of Salesforce AI in 2023, then joined Meta the following year to run its Business AI group. On social media she describes herself as a “conditional AI optimist,” a label that now reads like a warning she gave herself early.
“We Didn’t Know Whether to Cry or Celebrate”
In August 2025, as Meta’s Head of Business AI, Shih and her team built an agent that answered customer questions, remembered preferences and even offered discounts to close a sale. Owners adopted it fast because it kept shoppers engaged and sped up their sales process.
“It was everything you would want when you’re a builder in technology,” Shih recalls. Then came the feedback that changed her mind about what she had built.
“The feedback we got from the businesses using our software was that they no longer needed their staff,” she says. “We didn’t know whether we should cry or celebrate because we’ve been wanting it for so long and yet when we achieved our goal, we started to realize what the implications were.”
She described the same moment to Fortune in starker terms. “In that moment I knew that nothing would ever be the same,” she told the magazine. “You feel radicalized in that moment when you see it working.”
Most tech chief executives don’t talk about AI’s toll on jobs that plainly. Several have spent the past year softening their public language around AI job losses, even as their products keep automating the same entry-level tasks Shih watched disappear.
Recent Graduates Are Losing Ground Fast
Around the same time, Shih was hearing from younger relatives and recent graduates who couldn’t land work, some after job offers were rescinded. “I started looking into this problem more and what I found alarmed me,” she says. “We are in a youth unemployment crisis in this country.”
The Federal Reserve Bank of New York’s own labor market tracker backs her up. Recent graduates aged 22 to 27 face 5.6% unemployment, and 41.5% of them work jobs that don’t require the degree they earned.
“These are smart, talented, ambitious young people who should be gainfully employed,” Shih says. She argues the automation of entry-level tasks is not just a personal setback for graduates. It shrinks the tax base and the future voter base, she says, calling the fallout a societal cost, not just an individual one.
The numbers look worse next to other groups in the workforce.
| Group | Unemployment Rate |
|---|---|
| Recent college graduates, ages 22 to 27 | 5.6% (41.5% underemployed) |
| Young workers without a college degree | 7.8% |
| College-educated workers, ages 22 to 65 | 3.1% |
| All U.S. adults | 4.2% |
College graduates once had a clear labor market edge over their peers without degrees. That gap has narrowed sharply for the youngest cohort now entering the workforce.
Andrew Yang’s 2020 Warning Finds a New Believer
Shih didn’t build the New Work Foundation alone. Andrew Yang, whose 2020 presidential campaign centered on the job risks of automation and a universal basic income, joined as a founding adviser when the organization launched at the TIME100 Summit in April.
Yang spent that campaign being dismissed as a fringe candidate obsessed with robots taking jobs. Shih was, by her own account, one of the people who dismissed him.
When I first heard you talk years ago, Andrew, about the eventual impact of AI on jobs, I thought you were crazy. Now I feel like you were right.
Shih wrote that in an interview posted to Yang’s own newsletter, describing how rolling out Meta’s agents to more than a million businesses convinced her the same disruption was coming for the wider economy. Yang, for his part, has called the entry-level freeze “the canary in the coal mine for the broader economy.”
Small Business Hiring Is Actually Climbing
The picture isn’t uniformly bleak, and not every employer is behaving like the ones Shih watched cut staff. Witnesses before the House Small Business Committee this month argued the opposite trend is playing out among smaller firms.
Jordan Crenshaw, senior vice president at the U.S. Chamber of Commerce’s tracking of small business AI adoption, told lawmakers that 58% of small firms used generative AI last year, up from 40% in 2024 and 23% in 2023. Of those firms, 82% increased their headcounts over the past year, he said.
“These businesses are not using AI to replace people,” Crenshaw said. “They’re using it to help employees focus on high value work that requires judgment, creativity and relationships.” He also pressed Congress to set one national AI rulebook rather than a state-by-state patchwork, citing a think tank estimate that a fragmented approach could cost the economy roughly $1 trillion over a decade, with small businesses absorbing about $200 billion of that.
Other recent data complicates any tidy story. The Wall Street Journal reported that 27.3% of U.S. job seekers had been out of work six months or longer in June, near a five-year high, hitting white-collar workers and people in their prime working years hardest. Quartz has reported that some companies that cut entry-level roles are now rehiring that talent, a sign automation hasn’t been as clean as expected, and warned that gutting junior roles could eventually starve the pipeline of future managers.
Dear CC, JobClaw and Field Report Go Live
Shih’s answer to all of this is more AI, not less, aimed at graduates instead of employers. “The first problem we want to address is this information gap between people who work in AI and everybody else,” she says.
What started as a self-funded effort is now a nonprofit that accepts donations toward equipping young workers with the tools, information and skills to navigate an AI-reshaped economy. The organization runs on three connected initiatives.
- Dear CC – a media and data platform, styled as a modern update of the Dear Abby advice column, that goes role by role through common entry-level jobs in marketing, sales, engineering and accounting to explain how AI is changing each one
- JobClaw – an AI agent that matches a job seeker’s strengths and interests to open roles, surfaces skill gaps and points toward upskilling, working across job boards including LinkedIn and Indeed
- Field Report – a tool built to help recent graduates identify the strongest job options for their specific major
“We’re trying to create this marketplace where we can connect Gen Z job seekers with opportunities,” Shih explains. The foundation describes its approach on its official mission page as applying AI-native methods to both the entry-level freeze happening now and the longer-term reality of an economy being reshaped by agents.
The Next Data Check Comes in August
The New York Fed updates its graduate labor market figures quarterly, in February, May, August and November. The next reading lands in a few weeks, and it will say more about whether Shih’s bet is working than any pitch deck can.
Until then, the tension she named herself stands unresolved. The tool that made her want to cry and celebrate at once is the same category of tool she is now asking an entire generation to master faster than the businesses that stopped hiring them.
Frequently Asked Questions
What is the New Work Foundation?
It’s a nonprofit organization founded by Clara Shih in April to help young workers navigate an AI-reshaped economy, funded through donations after starting as a self-funded effort. Shih volunteers her time, and the foundation operates under the consumer-facing brand Dear CC.
How does Andrew Yang know Clara Shih?
Beyond his role as founding adviser to the New Work Foundation, Yang has interviewed Shih directly on his own podcast, where the two discussed what AI is already doing to entry-level white-collar jobs.
What does “underemployed” mean in the Fed’s graduate jobs data?
The Federal Reserve Bank of New York defines an underemployed graduate as someone working in a job that typically does not require a bachelor’s degree, even if they hold one. That distinguishes it from the unemployment rate, which only counts people actively looking for work and unable to find any job.
Which entry-level jobs are most exposed to AI right now?
Data compiled by the foundation’s own Dear CC platform points to software developers, financial analysts and accountants as among the most vulnerable roles. That exposure shows up in enrollment numbers too: software engineering enrollment dropped 8.1% in the 2025-2026 school year, according to the National Student Clearinghouse, and entry-level job postings have fallen 17% since 2019.
Are the New Work Foundation’s tools free to use?
Yes. JobClaw, for example, started as an open-source prototype posted publicly on GitHub, with the foundation planning a broader consumer release as the tools mature.
