E-commerce on Autopilot with AI

The problem

The client sells ring lights, wireless microphones, tripods, stabilisers and lighting kits to content creators, about 60 products across 7 categories, with nationwide cash-on-delivery. Growth in this category runs on a constant stream of fresh video ads on Facebook and Instagram, and every ad passed through several hands: someone picked the product, someone wrote the idea, an editor spent hours cutting the video and someone else uploaded it. Orders were confirmed by phone, figures lived in separate spreadsheets, and with no link between stock and advertising, money could go on promoting items that were out of stock.

Our approach

We did not hand the team another tool to learn. We built a system that does the work itself and leaves people with two small jobs: record a short product video and approve the result. We started with the biggest cost, ad production, recorded baseline sales and ad figures in week one so every result could be measured against reality, and kept approvals and budgets with the owner throughout.

How we delivered it

Set up (Week 1)

• Connected the online store and ad account
• Agreed the brand style
• Recorded starting sales and ad figures as the baseline

First ads (Weeks 1–2)

• Finished ads for the best-selling products
• Team approved the voice, look and tone

Whole catalogue (Weeks 3–4)

• Ads for every in-stock product
• Daily results tracking
• One-tap approvals

Full operation (Weeks 5–8)

• Product photos and social posting
• WhatsApp order confirmation
• Product research and the live dashboard

What we built

The team records a 20 to 40 second phone video of a product. The AI system picks products that are in stock, studies what has sold, writes several ad ideas, edits the video, adds local-language or English voiceovers, captions and price, and produces every size for Reels and Feed. The owner approves each ad on one screen, and the system places it in the ad account, tracks results daily, keeps what sells and drops what does not. Ads pause themselves when a product runs out. Around it, the same system produces studio-style product photos from a single phone picture, website images and banners, scheduled social posts for Instagram, Facebook and TikTok, WhatsApp order confirmation for cash-on-delivery, and a weekly shortlist of trending products worth stocking. A single dashboard brings sales, orders, ad spend and return, stock, ad approvals, the content calendar and product ideas together.

Results

Operating cost is down 30% and team productivity is up 50%. Ads, photos, social posts, order confirmation, product research and reporting now run without manual steps, one dashboard replaced the spreadsheets, and peak sale seasons can be covered in days instead of by hiring. Growth no longer means more editors and more computers; it means asking the system for more.

• 30% lower operating cost, measured against the pre-project baseline
• 50% higher team productivity with the same team
• Ads built around real product footage, so customers receive exactly what they saw
• Peak seasons covered in days, without hiring
• Knowledge about products and customers builds up in the system instead of leaving with staff