Ibrahim Alkhudairi

Theq: a two-sided marketplace for automotive services

Requirements from more than fifty automotive workshops, turned into a specified and prioritised MVP that is now live on the App Store.

Kind
Product & analysis
Role
Founder: product lead and main developer
Organisation
Theq
When
Aug 2025 – present
Duration
9+ months, ongoing
Tools
Workshop interviews, Functional specifications, Jira, App Store Connect
Stack
Flutter, Riverpod, Node.js, Express, TypeScript, PostgreSQL (Supabase), Firebase, Next.js, Python, FastAPI, LangChain, OpenAI API, Docker, Oracle Cloud
Links
App Store
  • Welcome
  • Home: recently viewed and popular workshops nearby
  • Workshop page with services, hours and booking
  • Book an appointment: car, services, photos and the problem
  • Map view
  • Browse by service
  • Search with service and city filters

Context

Theq is a two-sided marketplace for automotive services. Car owners post service requests and book; workshops receive and fulfil them. I started it in August 2025, and the first job was not code. It was finding out what workshops would actually use, which meant going to them.

My role

Founder, which in a five-person team meant owning the product end to end: requirements, scope, the data model and the sprint cadence. I made the prioritisation calls and carried the trade-offs between what workshops asked for and what a small team could ship.

Approach

Requirements came from more than fifty automotive workshops. Their input was qualitative, so I translated it into functional specifications and a prioritised MVP scope rather than a feature wish-list.

I designed and queried the PostgreSQL data model behind the platform on Supabase, covering users, workshops, service requests and booking transactions. Having the analyst also own the schema kept the specification and the database from drifting apart.

I also built most of the platform. The Flutter app serves both car owners and workshop owners, in Arabic and English. Behind it are an Express and TypeScript API on Supabase, Firebase for sign-in and push notifications, and a Next.js admin portal. The API and the AI service run as Docker containers on an Oracle Cloud VM behind Caddy, and GitHub Actions redeploys them on every push to main.

The app also has an AI assistant. Ask it in Arabic or English for a workshop and it searches the live listings; ask it for help and it takes you to the right screen. It is a LangChain tool-calling agent on OpenAI's gpt-4o-mini behind a FastAPI service, with unit tests in CI and a live evaluation suite of 58 cases covering search, navigation, refusals, language and multi-turn conversations. In its latest run every case passed three times out of three.

The team worked in Agile sprints for more than nine months. After launch on the App Store, adoption data replaced opinion as the input to prioritisation.

Deliverables

  • Functional specifications and a prioritised MVP scope derived from 50+ workshop conversations
  • PostgreSQL (Supabase) data model for users, workshops, service requests and bookings
  • The Flutter app for car owners and workshop owners, in Arabic and English
  • An Express and TypeScript API, a Next.js admin portal, and Docker deployment on Oracle Cloud
  • An AI assistant with a 58-case evaluation suite
  • Sprint plans and a groomed backlog for a five-member team
  • App Store release and post-launch iteration driven by adoption data

Outcome

Theq is live on the App Store, where it is rated 5.0 from nine ratings. The five-member team has shipped iteratively for more than nine months, and feature priorities are now set from usage rather than assumption.