An AI Freelance Marketplace Platform That Turns Site Audits Into One-Click Hires

Nine years with one client, and the most recent phase changed what the product is. Legiit went from listing gigs to telling a business what is wrong with its website and which seller can fix it.

Start your project

The build

Legiit is a services marketplace where businesses hire vetted freelancers for SEO, content and website work.

We worked with the Legiit team for roughly nine years, and the most recent phase changed what the product is. It went from a marketplace that lists gigs to a platform that tells you what is wrong with your website, what to do about it, and which seller can do it. At Aipxperts, we built the AI layer, the Command Center dashboard and the rebuilt marketplace front end that made that possible.

The problem

Growth work was split across tools that did not talk to each other. A business owner could hire a writer on the marketplace, then open Google Analytics in a second tab, Search Console in a third, and a one-off SEO checker in a fourth. Nothing connected the finding to the fix. The technical starting point made that harder to solve:

The marketplace was a conventional Laravel Blade and jQuery application with server-rendered pages, so any interactive dashboard meant fighting the existing rendering model.

There was no unified dashboard of any kind, and no AI layer to build one on.

Buyers had no way to see site issues, next actions and relevant services in one place, which meant the marketplace only captured demand that a buyer had already diagnosed themselves.

That last point is the commercial one. A marketplace that waits for buyers to arrive knowing what they need is a smaller marketplace than one that shows them.

What we built

We kept the Laravel core that already ran orders and payments, and built outward from it. The marketplace front end moved to React with TypeScript on Vite. The Command Center was built in Next.js. A separate Python service on FastAPI handles everything AI, and the PHP and JavaScript applications call it over REST.

Service marketplace

Listings, cart and checkout, orders, reviews, and paid boosts and sponsorships, rebuilt in React and TypeScript so seller and buyer flows respond without a page reload on every action.

Command Center

The dashboard a buyer lands on after connecting a site. It runs SSL, SEO, health and PageSpeed checks, groups the findings into readable insights, tracks rankings over time, and puts a hire button next to each problem it finds.

Website crawl pipeline

AWS Lambda crawlers feed results through Kafka consumers into grouped insight sets, so the audit runs asynchronously and the user sees results arrive rather than waiting on a spinner.

Marketplace chatbot with RAG

A chatbot that answers questions about orders, services and courses by retrieving from marketplace data held in Milvus, rather than guessing from a general model.

AI business tools

Brand voice and customer avatar generation, SWOT summaries, content clusters, social posts, cart recommendations and a review assistant, all served from the FastAPI layer.

Real-time chat and notifications

Buyer and seller messaging over Pusher with MongoDB behind it, so conversation history stays out of the transactional MySQL database.

AI listing and member review

Automated first-pass moderation of new services and members, which took manual approval off the operations team's queue.

Admin panel

Users, services, permissions and day-to-day operations, kept on Blade where server-rendered pages were the right tool.

Flutter mobile application

A companion app with push notifications through Firebase, so buyers and sellers can act on orders and messages away from the desktop.

The outcome

The measurable outcome here is structural rather than a single percentage. The buyer journey changed shape. That matters commercially because it moves Legiit from a marketplace people search when they already know what they want, to one that creates the demand it then fills. Every audit finding is a potential order with the reason for the order attached.

Before: hire talent in one tab, guess what to fix in another, track results nowhere.
After: connect a site, the platform flags issues and opportunities, and the buyer hires a vetted seller against a specific finding without leaving the product. Work and results then live in the same place.

Tools and Technologies

Laravel 10

React + TypeScript

Next.js

FastAPI

Flutter

Milvus

Kafka

MongoDB

Redis

AWS Lambda

Pusher

Firebase

Planning something similar?

Building a marketplace that needs an intelligence layer on top of it? Tell us what your users are trying to work out for themselves.

Start the conversation

Related Case Studies

All case studies

Marketplace

Keeping Marketplace Deals Moving After the Buyer Closes the Laptop

A services marketplace runs on conversation, and conversation stops when the buyer closes the laptop. Chat, order completion and web-login approval moved…

Read the build

Marketplace

Building the AI Layer Behind a Live Marketplace Without Touching Checkout

The marketplace was already live and taking payments, which ruled out rebuilding it. The AI layer runs as a separate FastAPI service,…