Building the human face of autonomous procurement AI at Zip - where half a trillion dollars of enterprise spend meets a five-person design team.
There is a moment in every agentic workflow where a machine has done everything it can and needs a human to take the next step. That handoff - the moment the interface either earns trust or destroys it - is Akash Talyan's domain.
Talyan is a product designer at Zip, the San Francisco-based AI procurement platform that raised $190 million at a $2.2 billion valuation in late 2024. His specific charge: design the agent experience - the layer of UI, feedback, and guidance that makes Zip's 50+ autonomous procurement AI agents legible, trustworthy, and actually useful to finance, legal, and procurement teams managing billions in spend.
Before this, he was redesigning integration marketplaces at Lucid Software (42% more integrations per month, documented), building multi-wallet flows at Coinbase, and designing participant experiences at Verily for the Verily Me consumer health app. Before California, he was a design engineer at Honda Cars India, working on physical products in a manufacturing environment that teaches you something software design almost never does: that how something feels to hold matters as much as how it looks.
I'd rather have off-the-shelf UI with great interaction design, a strong voice, and thoughtful guidance than a beautifully crafted UI with unclear directions or a confusing user journey.
- Akash Talyan, Product DesignerBefore Zip's AI agents, Talyan was at Georgia Tech's School of Industrial Design exploring how HDPE plastic bags could become raw material for product design exploration - and building "Kirisense," a system for making rigid materials bendable. The design instinct that turns waste into form is the same one that turns opaque AI workflows into clear user experiences.
Most designers plant a flag in one vertical. Talyan has moved across four - each time carrying hard-won domain knowledge into unfamiliar territory. That range is not accidental. It is the curriculum.
Each industry hands a designer a different set of constraints, user mental models, and stakes. Crypto wallets are high-trust, high-anxiety interfaces where one wrong tap costs real money. Healthcare apps handle data where every word of copy carries liability. Enterprise procurement involves approval chains, compliance requirements, and ERP integrations that would make most consumer designers quietly resign. Talyan has navigated all three before arriving at Zip's agentic challenge.
Relative domain depth across Talyan's documented career experience:
* Relative, based on documented work history
Most product design problems ask: how do users accomplish a task? Agent experience design asks a different question: how do users know whether to trust what an agent already did?
When an AI agent automatically routes a $2 million vendor contract renewal or flags a supplier for risk violations without being asked, the interface has to communicate what happened, why it happened, what level of confidence the system had, and what the user can do about it - all within the cognitive budget of someone managing 50 simultaneous procurement requests.
That requires not just visual design skills but a deep understanding of how people build and lose trust in automated systems - exactly the domain that Georgia Tech's HCI program is trained to produce.
Navigating ambiguity and developing user-centered solutions, transforming complex problems into clear, scalable systems that drive business outcomes.
- Akash Talyan, professional philosophy (LinkedIn)