Editor’s note: This article is based on insights from a podcast series. The views expressed in the podcast reflect the speakers’ perspectives and do not necessarily represent those of this publication. Readers are encouraged to explore the full podcast for additional context.
The public sector is racing to understand, regulate, and responsibly deploy AI—but few people have been at the center of this transformation like Shreya Amin, New York State’s first and former Chief AI Officer. In a recent episode of the “CAIO Connect” podcast, hosted by Sanjay Puri, Amin offers a rare, behind-the-scenes look at what it takes to bring AI into one of the largest and most complex government ecosystems in the world.
Amin’s career is anything but linear—and that’s her superpower. With roots in theoretical math and physics, she transitioned into data science roles in finance, software, edtech, and civic tech. This unique blend of analytical rigor and real-world experience eventually positioned her to become New York State’s first CAIO, where she led statewide strategy, governed responsible AI use, and spearheaded major data transformation efforts.
When Amin stepped into the role, New York already had an AI acceptable-use policy—but it was just a high-level document. Her job was to translate it into action.
READ: Viral Tripathi on building real-world AI impact at C1 (
Amin worked directly with agency leaders to help them understand how to implement the guidelines, not just read them.
She led a statewide AI training rollout for employees with vastly different levels of AI familiarity. The mission wasn’t just education—it was reassurance. Many employees feared job displacement, so the program emphasized safety, new opportunities, and hands-on practice through an enterprise-safe LLM tool.
Amin introduced something the government doesn’t often see: a structured product vision. She focused on high-volume pain points—like document backlogs—and created a sprint-based playbook for evaluating and launching new AI tools. The result? The very first AI product went live in just four weeks, giving agencies a template and a confidence boost.
Amin spent her first few months meeting with stakeholders across agencies. With long-tenured staff and union environments, she learned that reskilling can’t be rushed—it must unfold in phases.
In the public sector, value matters even more than in the private sector. Amin prioritized tasks that were:
- High volume
- Measuring-friendly
- Repetitive and time-consuming
If a task takes 15 minutes and AI can bring it down to two, that’s a real impact, she noted.
READ: ‘AI will change every role’: Tom Andriola on building enterprise AI that actually works (
Instead of endless paperwork, Amin pushed for embedded guardrails—like logging and privacy checks—so AI projects could move faster without increasing risk.
Since hiring AI experts is notoriously slow in government, she built small R&D groups with graduate students and forged advisory networks with prominent AI leaders.
Since hiring AI experts is notoriously slow in government, she built small R&D groups with graduate students and forged advisory networks with prominent AI leaders.
Amin believes autonomy should match the cost of error. Agents are great for coordination or suggestions, but not for regulated, high-stakes decisions—at least not yet. And while data quality is a challenge; the real issue is coordinating data across legacy systems, not a shortage of data itself.
Amin’s closing wisdom is powerful:
- Know the details and tie AI directly to business outcomes.
- Think like a product leader—understand the user’s world.
- Share playbooks so the next wave of CAIOs can build faster.
Her journey is a blueprint for how the government can embrace AI responsibly, boldly, and with a human-centered focus.