We sat down with Akylai Kasymkulova, Co-Founder and CTO of Human Delta, to reflect on building AI infrastructure, navigating the realities of early-stage startups, and solving one of the biggest challenges facing enterprise AI adoption today.
Originally from Kyrgyzstan, Akylai’s path into technology began with a strong academic foundation, graduating high school with the country’s highest academic distinction before studying computer science at the University of Illinois Urbana-Champaign. During her time there, she conducted large-scale AI and systems research at the National Center for Supercomputing Applications, working at the intersection of data, infrastructure, and machine learning. She later became one of the youngest Central Asian recipients of the 0-1 visa for individuals with extraordinary ability.
Akylai co-founded Human Delta with a clear observation: while AI capabilities continue to advance rapidly, most organizations still struggle with a more fundamental problem — their internal knowledge is fragmented, inconsistent, and difficult for AI systems to reliably operate on. Human Delta was built to address that gap by helping enterprises surface, structure, and validate institutional knowledge, transforming disconnected documentation and tribal knowledge into continuously maintained systems AI agents can depend on.
The company is backed by investors including Afore Capital and Susa Ventures, alongside leading operators and angels, and is already working with enterprise organizations such as Disney to deploy its platform in production environments. Across research, infrastructure, and company building, Akylai’s work has consistently focused on creating systems that make complex technology more reliable, usable, and scalable in practice.
To bring a more personal perspective to her journey, we invited Akylai to share a few reflections on entrepreneurship, AI, and building in a rapidly changing industry.
Reflections From the Journey
The messy part of building a startup is… You’re constantly deciding which chaos is signal and which chaos is just noise. A customer objection, a broken workflow, a weird enterprise process, or a team bottleneck can either be a distraction, or the exact clue that tells you where the company needs to go next.
One early mistake I’m glad I made was… Spending time close to customers before over-defining the product. It made the early roadmap messier, but it helped us see the real problem more clearly: enterprises do not just need more AI tools, they need the underlying knowledge foundation that makes those tools reliable.
What people assume is easy, but isn’t, is… Turning excitement into urgency. Everyone wants to talk about AI, but getting a large enterprise to actually change how it operates, buys, and deploys software takes a much deeper level of trust.
A belief about startups I’ve changed my mind on is… That speed only means shipping faster. I still think speed matters a lot, but I’ve learned that in enterprise, speed also means learning faster, narrowing faster, and being willing to throw away your own assumptions when the market tells you something different.
What keeps me going on harder days is… The gap between what AI can do in demos and what it can safely do inside large organizations is still enormous, and I think the companies that solve that gap will define the next decade of enterprise software and we aspire to be that company.
If I started again, I would… Talk to serious customers even earlier and be less afraid of the product being imperfect. Early on, it is tempting to wait until everything looks polished, but the best signal usually comes from putting something real in front of people with painful problems.
The question I’m asking myself most right now is… How do we become truly load-bearing for our customers? Not just a useful tool, but the layer of infrastructure they rely on as they deploy AI across more teams, workflows, and agents.
The reality of fundraising in the age of AI is… There’s more capital and more noise than ever, so differentiation isn’t optional; it has to be obvious very quickly. Investors see hundreds of AI companies, so the real challenge is proving that you are not just building another application, but attacking a structural problem in the market.
Continue the Conversation: Fundraising in Times of AI
Meet Akylai Kasymkulova at our upcoming Female Founder Roundtable on “Fundraising in Times of AI” on May 21 at Moon Creative Lab in Palo Alto. She’ll share how fundraising works in practice, common early mistakes, and how to build momentum when things feel uncertain.
🔐 Limited capacity — secure your spot early here.