Problem

Good Meta ads take expertise: picking audiences, writing creatives, setting budgets. Most small business owners don't have it, so they spend money on campaigns that never perform.

Outcome

One link in, a live campaign out. Owners launch and keep improving their ads without ever opening Meta Ads Manager.

How it works

1. A pasted URL feeds an analysis pipeline that reads the site and builds a brand profile: colors, tone of voice, audience and offers.
2. Durable workflows generate ad images and copy with Gemini and Claude, and every ad passes compliance checks before it reaches Meta.
3. A campaign planner turns the brand profile into ad sets, audiences, locations and a daily budget.
4. Daily cron jobs sync results from Meta, raise spend and fatigue alerts, and step the pixel goal up the funnel as data comes in.

Architecture

• Next.js App Router for the dashboard and its API.
• Supabase Postgres for brands, campaigns and results, with row-level security per workspace.
• Upstash Workflow for long-running, retryable generation and publishing jobs.
• Meta Marketing API for publishing and insights; Stripe for billing.

Problem

Leads arrived through the website, WhatsApp, calls and email, then lived in spreadsheets and chat threads. Follow-ups slipped through the cracks and every quote took hours.

Outcome

Every lead is assigned, followed up and quoted in minutes, in one system from first call to final invoice.

How it works

1. Website forms, WhatsApp, cloud calls and email all land in one shared lead inbox through webhooks.
2. Leads are assigned by role, with team visibility rules and real-time notifications.
3. Follow-ups, quotes and invoices all hang off the same lead record, so nothing falls between steps.
4. Quotes render as branded PDFs and go out by email or WhatsApp, with delivery tracking.

Architecture

• Next.js app and API covering sales, operations and HR.
• Supabase Postgres with role-based access for each team.
• Puppeteer renders PDF quotes and invoices; Resend sends email.
• WhatsApp Business API webhooks feed a shared team inbox.

Problem

Agents had nowhere central to list Dubai projects, and buyers couldn't search them by area, budget or payment plan.

Outcome

Agents list, admins approve, and buyers find the right project in one place, with smart filters, a mortgage calculator and an enquiry form.

How it works

1. Agents submit listings from a dashboard form with gallery uploads and a rich text editor.
2. Admins work through a review queue, approve or reject listings, and flag them as featured, high ROI or exclusive.
3. Buyers search by city and area, filtering by type, bedrooms, furnishing, handover date and payment plan.
4. Every property page adds a mortgage calculator, a brochure download and an enquiry form that feeds the dashboard.

Architecture

• Next.js App Router with server-rendered property pages and a generated sitemap for search engines.
• MongoDB with Mongoose models for properties, areas, developers and enquiries.
• Ant Design for the admin dashboard; Tailwind CSS for the public site.
• Cloudinary for images and floor plans; Resend for enquiry emails.

Problem

People who referred property buyers had no way to see where a lead stood or when they would be paid.

Outcome

Referrers follow every lead live and get a push notification the moment their commission is released.

How it works

1. The mobile app talks to a token-secured REST API for search, leads, profiles and payouts.
2. Each referral becomes a lead that moves through six stages, from submitted to commission released.
3. Every stage change in the admin panel sends a Firebase push notification to the referrer.
4. Releasing a commission updates the referrer's earnings and records the bank payout.

Architecture

• Next.js API routes serving the mobile app with token-based auth.
• Supabase Postgres for leads, referrers and payouts.
• Firebase Cloud Messaging for push notifications.
• Cloudinary for project images, plus a web admin panel for the sales team.