AI can write faster, build faster, and automate more than most business teams could have imagined a few years ago. But the more powerful the tools become, the more important the real question gets: should we use AI here at all?
That question runs through Anshika Agrawal’s work and thinking. A recent MBA graduate from San Jose State University, Anshika came to business school after studying computer science engineering in India and spending six years in tech and startup environments.
Her MBA experience sharpened that perspective, especially in the AI for Social Good course taught by Dr. Yu Chen. The course was not simply about using AI. It pushed students to think like consultants: what problem are we solving, what trade-offs are involved, and where should human judgment stay in the loop?
Building a companion, not a decision-maker
For her class project, Anshika’s team focused on the grant application process for the city of San Jose. Applicants often struggle to understand which grants they qualify for, what documents they need, and whether their application is strong enough. City reviewers, meanwhile, spend time sorting through incomplete or inconsistent submissions.
The obvious AI shortcut would have been to generate grant applications automatically. Anshika’s team took a different path. They built a platform to help applicants discover relevant grants, check whether their application was ready, identify missing information, and get guidance along the way.
The distinction mattered. As Anshika put it, the goal was not for AI to make decisions on behalf of applicants or reviewers. It was to act as a companion or support system. The tool could flag weak areas, summarize information, and help people think through the process, while leaving the final judgment with humans.
Because the team had limited access to city stakeholders, they also used AI-generated stakeholder simulations to challenge their assumptions. It was not a replacement for real interviews, but it helped them think through frustrations and gaps from both the applicant and reviewer side.
Vibe Coding lowers the barrier, but not the responsibility
Anshika has also become a hands-on builder with tools like Replit and Lovable. Her first vibe-coding project, Pantry Pal, helped people reduce food waste by recommending recipes based on what was already in their fridge. Since then, she has used AI tools to prototype personal productivity ideas and business concepts.
What excites her is not just the speed. It is the fact that people without traditional coding backgrounds can now test ideas in hours. In her class hackathon, she saw non-technical teammates realize they could build a website or prototype by describing what they wanted.
But that accessibility does not remove the need for judgment. One story from the course stuck with her: a factory trying to solve the problem of empty bottles being shipped considered sophisticated technology, until someone suggested putting a fan by the conveyor belt to blow the lighter empty bottles away. The lesson was simple. Businesses can rush toward complex AI before understanding the actual problem. AI may not be the solution for everything.
For Anshika, that is the heart of responsible AI in business. Automation can remove repetitive work and improve efficiency, but it should not quietly replace accountability. In situations involving trust, privacy, or personal impact, humans need to retain decision-making authority.
The future of AI in business may not belong to the people who automate the most. It may belong to the people who know exactly where automation should stop.