How We Built Julex 24 in 3 Months: From Instagram DMs to Custom Checkout

Before Julex 24 had a website, it had an Instagram account and a spreadsheet. Orders came in as DMs. The founder would screenshot a ring, paste a price, and send a UPI QR code. It worked — until it didn't.

By the time we got on a call, she was losing an hour a day to order confusion. Two customers had paid for the same piece. Another had paid and never received a confirmation. The business was real; the infrastructure was held together with tape.

Twelve weeks later, julex.shop went live. Here is how the three months actually went.

Week 1: The working session, not the pitch

We do not pitch on the first call. We sit down and walk through the business. For Julex, the real questions were not "what pages do you want" — they were:

• Are pieces one-of-one or do you restock?
• Do you want customers to see stock counts or just "sold out"?
• Who packs the order? Where does that notification need to go?
• What happens on a return? (Answer for jewelry: almost never, but you still need the flow.)

The answers to those four questions shaped the entire data model. One-of-one inventory meant every piece was its own SKU, not a variant. Manual packing meant we needed a simple admin order queue, not a full WMS. That cut two weeks off the build right there.

Weeks 2-4: The boring foundation

We shipped the stuff nobody sees first. Next.js 15 App Router, Prisma on PostgreSQL, auth, admin-protected routes, S3-backed image uploads, a shadcn-based admin UI that the founder could actually use without training.

A Shopify store would have skipped this. We built it on purpose because Julex needed workflows Shopify's templates fight against — custom engraving fields at checkout, a waitlist for restocked pieces, and a founder's note that shows up in the order confirmation email.

If your store is six T-shirts and a mug, use Shopify. If your product has a story and a workflow, custom pays for itself in the first year.

Weeks 5-8: The storefront

This is where most agencies start and most agencies stop. We started here in week five on purpose — by then the admin tools worked, so the founder was uploading real products while we built the public pages. Day-one content, not stock photos.

We shipped:

• Product pages with 360-degree image galleries using Embla
• A filter-and-sort catalog that stays fast even with 200+ SKUs
• A checkout that handles Indian address formats, GST, and Razorpay without redirecting off-site
• Framer Motion page transitions that feel expensive without being slow

If you are a founder reading this and thinking "I want the same thing but for my category" — that is usually what starts the conversation. Tell us what you're building here.

Weeks 9-11: The ugly middle

Every three-month build has an ugly middle. For Julex it was shipping integration. Courier APIs are inconsistent; one vendor's weight unit is grams, the next is kilograms, a third lies about serviceability. We spent a week on what we had estimated at two days.

This is the part nobody writes case studies about. It is also why 2-3 months is the real timeline and 3 weeks is a fantasy. If a developer quotes you 3 weeks for an ecommerce build, they are going to skip this week — and you will find out about it when the first complaint email arrives.

Week 12: Launch, and what we'd do differently

Julex went live on the date we quoted in week one. First organic order came in 36 hours later. No tape. No screenshots. No duplicate payments.

If we were starting this build again, two things would change:

1. Ship the admin in week one, not week three. The founder was uploading products into Notion while we built the CMS. That context switch cost both of us time.
2. Write the email copy before the email templates. We built beautiful transactional emails and then had to rewrite them twice because the founder's voice is warmer than our placeholder copy.

Neither of those is a reason not to do the project — they are reasons to do the next one better. That is the real reason we share case studies. Every project teaches the next one.

The takeaway for founders

If you are running a business out of DMs and a spreadsheet, a real product is closer than you think. Three months. Two people. A stack you own. See more of what we have shipped, and if you want yours on that list, start a conversation here.