A Watch Trader's Offers, Counters and Chaos, Moved Into One System
A luxury watch business that ran across notepads, inboxes, WhatsApp, Telegram and a label tool. First a site with live inventory and a make-an-offer flow, live in about a month. Then a custom CRM in 8 weeks and AI automations in 4 that sort every inquiry and draft every reply, with a person approving each one.
- Systems Shipped
- 50+
- Client Retention
- 98%
- Uptime
- 99%
- Engineers & Designers
- 40+

The Obstacle
A watch trader buys from individuals who want to sell a watch and sells to retailers who serve the end buyer. The business runs on getting the purchase right, which means knowing the market, moving fast when a good piece appears, and negotiating well, and on keeping track of everything in between: who's selling, who's buying, what shipped, whether the label was made, who still owes what.
The day ran across notepads, email inboxes, WhatsApp, Telegram and a label tool, with the owner hopping between them. An offer would arrive by WhatsApp, a counter would go out by email, the shipment would be tracked in the label tool, and the record of the whole deal lived in the owner's head. It worked because the owner is good at it. It couldn't grow, because nobody else could do it.
They came to us for marketing: more individual sellers, a bigger retailer network. We said the system had to come first, because more sellers and more retailers pouring into notepads and chat threads would only make the day harder.
Discovery: Following One Watch
We mapped the business by following a single watch from the moment an individual offers it for sale to the moment a retailer pays for it. The map had a buy side and a sell side, and the two were handled in completely different tools with nothing connecting them. The negotiation, the heart of the business, happened in whichever chat app the counterparty preferred, and the record of it disappeared when the thread scrolled.
Three things came out of discovery that shaped the build. First, offers and counter-offers are the business, so they had to live inside the system. Second, a watch comes in three ways, bought outright, taken as a trade-in or consigned, and each one settles differently. Third, payment is by wire; there is no payment gateway and no checkout in the usual sense. A template store assumes a cart and a card. This business has neither.
What We Built
- First, the site. A website the product deserved, with the inventory live and a make-an-offer system: a buyer bids on a watch, the trader counters, and the back-and-forth happens inside the site instead of across chat threads. Every offer has a record. It went live in about a month.
- Then, the CRM. One screen for the whole business, built in 8 weeks. Every seller inquiry, from the site's free-offer form, email or Telegram, lands in one inbox, linked to the right customer and watch, and is filed as sell outright, trade-in or consignment, each with its own steps through to settlement or payout. Stock, WooCommerce orders and buyers' Make an Offer bids are managed without opening the WordPress admin. When a client wants a specific watch, the team sources it through the wholesaler and dealer network, and grows that network with map-based searches, reseller sign-up approvals and new-stock announcements to approved resellers in a few clicks. Every customer and watch has one record with a full history, every status change is logged by name and time, and admins decide who can see and change what.
- Last, the AI automations. Built in 4 weeks on top of the CRM. When a message arrives by email, website form, WhatsApp or Telegram, an AI workflow filters out spam, works out the watch, its condition, box and papers, and the customer's budget, checks stock and drafts the reply. If the watch isn't in stock, it drafts a sourcing email to a supplier and turns the supplier's answer into a draft offer. Background jobs check the inbox every minute, keep orders and stock in sync with the website, and run scheduled follow-ups.
The AI Automations: Drafts, Not Decisions
The AI does the typing; people keep the decisions. It never sends a message, broadcasts to resellers or commits to an offer on its own. Anything unclear or low-confidence goes straight to a person, every draft waits for a CRM user to approve and send it, and the CRM records who approved each message and when. The system learns from the replies the team actually sends, so the drafts get closer to their voice over time. The order of the build mattered too: the site first, because it moved offers out of chat threads on day one; the CRM next, designed around real offer data already flowing from the site; the AI last, because it can only draft a good reply when the stock, the customer and the history are already in one place.
What the Trader Owns
The site, the hosting and the domain are in the trader's name, and the CRM and the AI automations are deployed the same way: on their own infrastructure, with the code, the data and the prompts handed over. A small team is the one that can least afford to depend on a vendor, and the system is built so it doesn't.
What Changed
- Day-to-day operations are organized and run from one screen, in the owner's own feedback
- Every inquiry, from four channels, lands in one inbox linked to the right customer and watch
- Offers and counter-offers happen in one place, with a record of every step
- Every action is logged by name and time, and admins decide who sees and changes what
- Replies are drafted by AI and sent by a person, never the other way round
What a Deal Looks Like Now
A seller fills in the free-offer form on the site. Before anyone opens the inbox, the AI has identified the watch, its condition and papers, checked stock and drafted a reply, and it is waiting with the customer's full history beside it. A team member reads it, adjusts the offer if needed, and sends it; the CRM records who approved it and when. If the seller counters, the counter lands on the same record. On the selling side, a buyer's bid on the site, the WooCommerce order and the wire payment all sit with the watch, so the question "what happened with that piece" has one answer in one place.
What We'd Tell Another Founder
When a client asks for marketing and their operation runs on notepads, marketing first just fills the notepads faster. Organize the operation, then pour demand into it.
The second lesson is about templates. The moment a business has a non-standard flow, offers instead of prices, trade-ins and consignments instead of simple purchases, wire instead of card, the template is working against it. Build around the flow, then let the AI do the typing while people keep the decisions.
They mapped how our business actually worked before they built anything. The system fits us; we didn't have to fit the system.
Tell Us How Your Operation Runs.
We'll Tell You What to Build.
A free 30-minute consultation with our founder. You leave with a clear view of what to build, what to automate and what to leave alone. No pitch deck. It is the shortest way to find out whether we are the right custom software development company for your operation, and we'll say so either way.
Book a Free ConsultationFive initial discovery calls a week, one a day, 11 AM to 3 PM Eastern.