Ikram Ul Haq
/now

What I’m
up to now

Last updated

A /now page — what has my attention at this moment, not a changelog.

Most of my attention is on Kendra — the AI stylist I lead at WearThat — and on the agent-driven workflows I use to build it.

01

Spotlight

Kendra

WearThatLead, AI stylist

Kendra is an AI stylist. I lead its creation. It runs in two places at once — behind the scenes for WearThat's in-house stylists, and directly in the hands of customers — and the same styling intelligence has to serve both, despite them wanting very different things from it.

Facet A · 01 of 02

Julia — the stylist portal

WearThat's in-house stylists work in a web portal called Julia. Kendra curates outfits for a customer automatically, and the stylist vets them rather than building each look from nothing. It shifts the stylist's job from assembly to judgement.

  • Auto-curates complete outfits per customer, which the stylist approves, adjusts or rejects.
  • Surfaces inspiration and suggestions for styling a particular customer, drawn from their past purchases, likes and dislikes, and size and fit preferences.
  • Keeps a human in the loop by design — Kendra proposes, the stylist decides.
The Julia stylist portal, showing Kendra-suggested pieces being assembled into a look for a customer
Box styling in Julia. Kendra's suggestions carry its own tag, the stylist builds the look from them, and the customer's sizes and fit preferences sit across the top so every suggestion is already filtered to what will actually fit.
Kendra AI assistant panel answering a stylist's question about a client
Stylists can also just ask. Kendra answers from the customer's style profile and returns pieces inline.
Facet B · 02 of 02

The customer app

In the customer-facing app, a customer talks to Kendra directly. The same intelligence that assists a professional stylist has to hold a conversation with someone who has never used a stylist.

  • Get styled for a specific occasion, from day-to-day to an event.
  • Get a full colour analysis that works out which colours suit them.
  • Get styling ideas for the wardrobe they already own — photograph one piece, get new ways to wear it.
  • Find the closest matching pieces to an outfit they saw somewhere else on the internet.
8 screens in the gallery
Kendra's introduction screen in the WearThat app
The customer's entry point into Kendra.
Ask Kendra screen listing four styling capabilities
Four ways in — restyle a piece you own, get styled for an occasion, find pieces you'll love, or get a colour palette.
Occasion selection screen for personalised styling
Occasion framing, before any recommendation is made.
Kendra conversation recommending workwear with shoppable products
A styling conversation resolving to real, shoppable pieces in the customer's size.
Generated style guide showing one outfit worn three ways
A generated style guide — one brief, three ways to wear it.
Restyle what you already own — photograph a piece, get three outfits
Photograph one piece you own; get three new outfits back.
Show Kendra a photo and she finds the closest match
Visual search — upload a look from anywhere, get the closest match from the edit.
Weekly AI-curated mood board in the WearThat app
A weekly mood board, curated per customer.
02

Also

Studying

An Executive Master's in Artificial Intelligence at LUMS, on weekends.

Exploring
  • Agent-driven development workflows — Claude Code, Antigravity, Codex and OpenCode, and what actually holds up in production.
  • This site, as a place to write things down.