AI-assisted coding is reshaping how coding is taught, learned, and applied in practice. For years, Google Colab has empowered students, developers, and researchers with a freely accessible, cloud-hosted Jupyter Notebook environment right in their browser. However, Google software engineers noticed that a variety of users were encountering different challenges and needed more tailored AI assistance. They invited students in MIT Sloan’s Corporate Entrepreneurship Lab (CE-Lab) to help them explore possible solutions.
“Despite Google’s scale, the Colab team operates with the agility of a startup, open to new ideas, quick to iterate, and deeply committed to solving real user problems,” notes MIT Sloan Fellows MBA student Sasa Phanitsombat, SFMBA ’25. “Their focus on user-centric innovation is something they actively practice.”
With this in mind, the MIT team set out to uncover key pain points students and educators of all experience levels face when using Colab and recommend improvements that would align with their workflows and learning processes. The team’s host was Google Software Engineer and MIT Alum George Whitfield, who is also an MIT Sloan lecturer and Entrepreneur in Residence at the Martin Trust Center for MIT Entrepreneurship. With his guidance, the students quickly formed hypotheses about problems the product could solve—which they systematically tested in a series of one-on-one customer interviews.
“With his deep ties to the Trust Center, George struck a rare balance, offering just enough guidance, while trusting us with the freedom to explore and deliver unique insights,” says MIT Executive MBA student David Kawesi-Mukooza, EMBA ’26. “That trust made our work not only relevant, but genuinely impactful.”
Theory into action
The MIT team’s findings about users— from beginner to advanced—uncovered when and where AI assistance is most effective and informed strategies to enhance adoption, improve usability, and tailor features that support students' coding progression. In parallel, they studied how educators and teaching assistants introduce coding in academic environments. This included addressing concerns around academic integrity, student over-reliance on automation, and challenges in assessing learning outcomes when AI is involved.
“This project was a unique opportunity to drive real impact at the intersection of AI and education,” says Prashasti Agrawal, SFMBA ’25. “Collaborating with the Google Colab team allowed us to apply user-centric thinking to a product that reaches millions of learners globally. Seeing our research directly influence product development was incredibly fulfilling—and perfectly embodied the spirit of MIT’s Action Learning labs: learning by doing, with tangible outcomes.”
For Kawesi-Mukooza, CE-Lab was one of the most meaningful experiences of his time at MIT. “Coming off Entrepreneurial Strategy and Disciplined Entrepreneurship, this course brought those lessons to life in a real-world setting—showcasing how the resources and stability of an established organization can amplify innovation when paired with entrepreneurial thinking.” He adds, “This course perfectly captured MIT’s ‘mind and hand’ philosophy—turning theory into action and making learning deeply tangible.”
Fellow team member Rafael Maldonado, SFMBA ’25, agrees. “It was a true 360º learning opportunity, from exploring a real-world problem to thinking critically about user needs and how AI can support them.”
Insights into action
According to Whitfield, the timing of the project was perfect, as his team at Google was already building a product to address pain points that the CE-Lab students also identified.
On May 20, 2025, shortly after the students completed their project, Whitfield’s team at Google I/O announced the launch of a reimagined AI-first experience for Google Colab.
“Now,” Whitfield says, “Colab makes it easy for coders of all skill levels to get help from an AI agent to both iterate quickly while prototyping and debugging and dive deep when needed into advanced architectures and analysis.”
