Built the design foundation at the Series A AI startup, supporting enterprise customer wins, including Fortune 500 logos, and Gartner Magic Quadrant recognition.

Built the design foundation at the Series A AI startup, supporting enterprise customer wins, including Fortune 500 logos, and Gartner Magic Quadrant recognition.

Role

Founding Designer

Timeline

Mar 2019 - Feb 2020

Core team

Core team

Problem

The product worked. The interface assumed every user could write code.

The product worked. The interface assumed every user could write code.

The product worked. The interface assumed every user could write code.

Saturday afternoon at Kohl's, March 2019. Marketing in Chicago notices revenue is below target. They call Kohl's tech in Milpitas. Tech calls ZineOne's CPO. A real-time offer takes two hours to go live.
That was the product I was hired to design for. I joined Session AI (then ZineOne) in March 2019 as the founding designer. Series A, a thirty-person team including three founders, $2.5M raised. The ICE platform had been live at major banks and retailers since 2017, with forty-plus use cases at Kohl's and twenty-two million users at HDFC Bank. The technical foundation was solid.
The interface was built by engineers, for engineers. Configuring a campaign meant editing JSON. Marketing managers, the buyers signing the checks, couldn't run the product without engineering.
Takeaway
The technical foundation was solid. The interface assumed users it didn't have.
Reframe

The brief was UI polish. The work was building the design foundation from zero.

The brief was UI polish. The work was building the design foundation from zero.

The brief was UI polish. The work was building the design foundation from zero.

There was no design team before me. No design system. No shared visual language across the product, the pitch deck, and the demo prototypes. Each surface looked like a different company.
UI polish on the platform alone would have moved one surface forward and left the rest behind. The work was scoping the role within the broader gap: design system, product UI, pitch design, demo prototypes, and research practice.
Takeaway
Founding design is not drawing screens faster. It is deciding what the design function is for.

Data Collection

Data Collection

Data Normalization

Data Normalization

Data Decisioning

Data Decisioning

Data Enrichment

Data Enrichment
Intelligent Customer Engagement
Intelligent Customer Engagement

Dynamic
Segments

Dynamic
Segments
Scheduled
Actions
Scheduled
Actions
Triggered
Actions
Triggered
Actions
Real-time Recommendations
Real-time Recommendations
Personalized Customer Experiences
Personalized Customer Experiences
Approach

The product was built for engineers. It needed to work for marketers.

The product was built for engineers. It needed to work for marketers.

The product was built for engineers. It needed to work for marketers.

ZineOne's customers had two kinds of users on the platform. Solutions engineers configured campaigns. Marketing managers ran them. The interface served the first group well, but not the second at all.
The design move was to add a layer, not replace one. Manish Malhotra, Arnab Mukherjee, and Kendra Chen were the partners on product and design direction. A no-code surface for marketing users sat above the platform's full configurability. Engineers kept everything they had. Marketers got a way in.
This decision lets a marketer or campaign operator launch a real-time offer without engineering. Every intervention that followed was a version of it.
Takeaway
A founding designer's first job is choosing what the design abstracts and what it leaves alone.
Logo Wins

Enterprise wins at Kohl's, HDFC Bank, Walmart, and Williams Sonoma. Pilots at Macy's and L'Oréal.

Enterprise wins at Kohl's, HDFC Bank, Walmart, and Williams Sonoma. Pilots at Macy's and L'Oréal.

Enterprise wins at Kohl's, HDFC Bank, Walmart, and Williams Sonoma. Pilots at Macy's and L'Oréal.

Active deployments. Four enterprise customers were running on the platform during my tenure. Kohl's was the flagship, with forty-plus use cases across web, mobile, and store. HDFC Bank ran on the platform for net banking, mobile banking, and wallet services, with 22 million users. Walmart and Williams Sonoma had ML models deployed.

Walmart targets discount shoppers while Williams Sonoma targets higher-end shoppers, each brand has its own ML model.

Debjani Deb

CEO, Session AI · MarTech Series, March 2020

Walmart targets discount shoppers while Williams Sonoma targets higher-end shoppers, each brand has its own ML model.

Debjani Deb

CEO, Session AI · MarTech Series, March 2020
Pilot deals closed. Two more pilots closed during my tenure. Macy's signed in late 2019 to address cart abandonment. L'Oréal and Kiehl's also signed in 2019 after a team-built demo.
Seth Jennings, Senior Director of Sales Engineering, wrote in an internal email afterward: "A lot of work went into creating the simulated RTO dashboard as well as the ML artifacts mocks." The demo was well received.
The design did not close these accounts on its own. Sales closed them. The platform earned them. The design was the surface on which every conversation ran.
Takeaway
Founding designer work shows up as the surface that lets the company sell.
Design

Five targeted design interventions. Each one removed engineering as a dependency for a customer-facing decision.

Five targeted design interventions. Each one removed engineering as a dependency for a customer-facing decision.

Five targeted design interventions. Each one removed engineering as a dependency for a customer-facing decision.

Intervention 1

Made campaign configuration usable without engineering.

Designed a modular workflow surface so business users could configure campaigns end-to-end. The platform's full configurability remained accessible to solutions engineers.

Intervention 2

Made campaign performance readable without analyst support.

Designed the in-product reporting so marketing leads could read campaign performance without an analyst translating the numbers.

Intervention 3

Let marketers launch and manage campaigns without being at a desk.

Designed the iOS app for in-store and weekend decisions. This is the surface that responds directly to the Saturday-afternoon Kohl's scenario. A marketing manager in Chicago can launch, monitor, and pause an RTO from her phone without filing an engineering ticket.

Intervention 4

Created a shared system so the product and pitch spoke the same language.

Built the components, patterns, and visual language from zero. Twenty-five-plus screens shipped across ICE, the RTO dashboards, and the mobile app, all running on it. The pitch decks ran on it as well.

Intervention 5

Established how design worked across product, engineering, and sales.

Introduced the working patterns design needed across functions, from zero: handoff and review with product and engineering, implementation and QA partnership with dev teams, customer-facing collaboration with sales. Founding designer work meant building the operating model alongside the product.

Recognition

Debuted in Gartner's 2020 Magic Quadrant for Personalization Engines.

Debuted in Gartner's 2020 Magic Quadrant for Personalization Engines.

Debuted in Gartner's 2020 Magic Quadrant for Personalization Engines.

ZineOne was one of thirteen vendors evaluated in the report. It was placed in the Niche Players quadrant. Dynamic Yield, Salesforce, and Adobe occupied the Leaders quadrant.
For a Series A startup with around thirty people, debuting in the Magic Quadrant alongside enterprise platforms with hundreds or thousands of employees is the recognition. This reflects the stage.
Takeaway
Niche Players at Series A is the credential.
Learning

Founding design is building the function, not just the product.

Founding design is building the function, not just the product.

Founding design is building the function, not just the product.

The same person who designs the product UI also designs the pitch deck, the demo prototype, and the design system that runs underneath them. Each is a different instrument. The job is to decide which one matters this week and ship it.
The work that mattered most at Session AI was not the prettiest screen. It was the surface that let non-engineers run an engineering-grade platform. The founding designer's work is about removing dependencies.