Sheila Gulati

Sheila Gulati is co-founder and managing director of Tola Capital, the enterprise-software investment firm she established in 2010 after helping develop Microsoft’s cloud strategy and launch Microsoft Azure. Her career spans two major platform shifts: cloud computing and the emergence of AI-enabled enterprise applications.

Gulati grew up working in her family’s enterprise-software business, gaining experience in testing, documentation, marketing, and shipping. After studying science, technology, and international affairs at Georgetown University, she joined Robertson Stephens, where she worked with networking and software companies, including on public offerings for Juniper Networks and F5.

At Microsoft, she held corporate-development and developer-business leadership roles, worked on investments, acquisitions, databases, and developer tools, and co-led the company’s enterprise strategy during Azure’s launch. She subsequently founded Tola to back enterprise applications built on cloud infrastructure; its investments have included OSIsoft, hybris, Clipchamp, Pulumi, Klarity, Simpplr, and Nooks.

In 2023, Tola closed Tola Capital Fund III, a $230 million fund focused on early-stage AI-enabled enterprise software. Gulati sees a familiar concentration of infrastructure spending alongside a new proliferation of open-source models, specialized applications, and companies building governance, security, and evaluation into enterprise AI.

What she believes enterprise AI must get right

• Dynamic evaluation datasets: Static benchmarks can reward memorization without establishing real understanding. Gulati favors changing datasets and realistic domain-specific tests that reveal whether models can reason beyond familiar exam questions.

• Evaluation grounded in human values: Autonomous systems should be assessed against safety, context, user experience, and the actual goals they serve. She also questions whether personalized agents should reinforce users’ existing beliefs or challenge them, warning that unchecked personalization could deepen familiar digital echo chambers. Her AI Engineer World’s Fair conversation examines these questions alongside the challenges of interpreting black-box models.

• Useful AI without productivity overhead: In her essay about investing in Tana, Gulati describes software that reduces organizational friction instead of creating another system to maintain. Tana’s contextual knowledge graph connects people, information, and commitments to make AI assistance more responsive to how individuals actually work.

Gulati also recruits experienced operators to support portfolio founders and advocates expanding women’s participation in private investing. She founded Pratham USA’s Seattle chapter and serves on the education nonprofit’s national board.