Case study · AI in production

Fine Art Market Intelligence and Subscription Platform

A dedicated WeyBee team builds and modernises a global art-market data platform: 4M+ users, 300,000+ personalised alerts a day, 54+ microservices, a new dealer product and an AI market-insights chat.

Product engineeringLegacy modernisationMicroservicesAI integrationNew product build
IndustryArt and collectibles
EngagementDedicated product team
ServicesProduct engineering · Legacy modernisation · Microservices
Platform.NET MVC · .NET Core · Python
4M+users
300K+personalised alerts a day
54+microservices in 9 domains
The client

The client runs one of the art world's longest-established market-data platforms. It tracks auction results, works for sale, artist histories and exhibitions across more than 950,000 artists and tens of thousands of auction houses, galleries and venues. Collectors, advisors, dealers, galleries and auction houses rely on it daily, on a freemium subscription model.

The challenge
  • Keep a very large, data-heavy platform growing while its user base passed 4 million.
  • Replace a legacy alert system that could no longer handle hundreds of thousands of personalised emails a day.
  • Give the internal team one reliable system to manage artists, artworks, auctions, imports and moderation.
  • Launch new revenue lines, including a self-service product for art dealers and AI-powered market insights, without disrupting the core platform.
What WeyBee did

WeyBee provides a dedicated team that works as part of the client's product organisation. The team works across a .NET, Python and Next.js stack on AWS. It has delivered five major streams: the core platform, the internal content management system, a rebuilt alert engine, a completely new dealer product, and an AI chat for premium subscribers. Over time the platform moved to 54+ microservices across 9 business domains.

What it does
Market intelligence platform

Search millions of artworks, follow artists, view price trends and manage personal collections, with premium data gated by subscription tier.

Internal content management

The system the operations team uses daily for artists, artworks, auctions, accounts, campaigns, bulk imports, moderation queues and deduplication.

Personalised alert engine

A rebuilt, scalable pipeline sending 300,000+ daily emails, with timezone-aware delivery, tier-based content rules, user filters and an operations dashboard.

Dealer self-service product

A new end-to-end product: public dealer presence, inventory and listing management, image uploads and buyer-inquiry leads, with its own admin CMS, authentication and deployment pipeline.

AI market-insights chat

Premium subscribers ask questions in plain language and get structured answers with artist overviews, price charts and live auction data. Every session is logged for cost and quality review.

Payments and analytics

Stripe and PayPal subscriptions, engagement analytics and SSO for internal staff.

AI Engineering

An AI market-insights
chat for premium subscribers

Premium subscribers ask about artists and prices in plain language and get structured answers with charts and live auction data. We built the chat as a product feature, with cost and quality tracked from the first session.

Answers grounded in live data

Each answer combines an artist overview, price charts and current auction lots from the platform's own database.

Model-agnostic by design

LLM providers sit behind an abstraction layer, so models can be switched or combined without touching the product.

Cost and quality per session

Every session is logged with its running cost, so the feature stays profitable as usage grows.

Gated by subscription tier

The chat is part of the premium plan, turning AI into a reason to upgrade rather than a cost centre.

Explore AI Engineering
MARKET DATASources unified
LLM LAYERModel in production
INSIGHTSignals surfaced
CHATInside the product
UPGRADEWork done
How it is built
  1. 54+ microservices across 9 business domains, with a GraphQL data-aggregation layer.
  2. .NET MVC and .NET Core services, Python services, and a Next.js front end.
  3. PostgreSQL, MS SQL and OpenSearch for data and search; Redis for caching.
  4. AWS ECS, SQS, SNS, RDS, S3, CloudFront and Secrets Manager; email through SES and SMTP.
  5. LLM providers connected through an abstraction layer, with per-session cost tracking.
Results
  • A platform serving more than 4 million users.
  • More than 300,000 personalised alerts delivered every day on the rebuilt engine.
  • Two new revenue products launched: a dealer self-service portal and an AI insights chat.
  • Continuous delivery by a stable WeyBee team.
Technology
.NET MVC.NET CorePythonGraphQLNext.jsPostgreSQLMS SQLOpenSearchRedisAWS ECSSQS / SNSRDSS3 / CloudFrontStripe / PayPalLLM APIs
Next case study

AI Clinical Genetics Platform with Genetic Test Ordering

Start a project

Planning something similar?

Tell us what you're building and we'll suggest the right team, engagement model and first AI use case.

Ask AiBee

Let’s Connect

We reply within one working day.

What would you like to discuss?
4.6 Clutch 4.5 GlassdoorNDA on request100+ clients