Case Study | AI-Powered SaaS Training Marketplace for Enterprise Buyers
AI-Powered SaaS Marketplace for Global Training & Enterprise Buyers
EdTechAI / MLSaaS MarketplaceAWS
FiftyFive built a scalable, AI-driven training marketplace that matches enterprise buyers with providers through natural-language search. The platform handles real-time availability, ML-based course ranking, and GDPR-compliant enterprise access at global scale.
The client is a UK-based EdTech company operating in the corporate training sector. They approached FiftyFive Technologies after validating an MVP that proved clear market demand but lacked scalability, AI intelligence, and enterprise-grade architecture. Their goal was a transparent global marketplace connecting training providers directly with enterprise buyers, supported by AI-driven search and real-time availability across UK, EU, and Middle East markets.
Challenges
The Challenges That Started It All
Turning a validated MVP into a production-grade, two-sided SaaS marketplace meant solving for AI search accuracy, real-time data at scale, and enterprise compliance simultaneously. Every architectural decision had to hold under global traffic while a full build ran to a fixed nine-month window.
The platform had to attract and activate two very different user groups at once — training providers publishing catalogues and enterprise buyers sourcing courses. Each side needed distinct onboarding paths, dashboards, and engagement mechanics without one side's experience compromising the other's.
Generic search engines return poor results for corporate training queries, where intent is expressed in business language rather than course taxonomy. The system needed a domain-tuned NLP pipeline capable of interpreting free-form buyer queries and mapping them accurately to provider catalogues.
Course availability changes continuously, and stale data undermines buyer trust. The platform had to surface live availability and pricing while remaining responsive under heavy concurrent load, requiring a caching and indexing strategy designed for real-time consistency at scale.
Provider catalogues arrive in inconsistent formats, with the same course frequently listed multiple times under different titles. The platform needed to normalise metadata, deduplicate near-identical listings, rank results by relevance, and layer ML recommendations on top of the cleaned data.
Serving enterprise buyers across UK and EU markets meant meeting GDPR obligations on data handling, consent, and retention from day one. Authentication and authorisation also had to satisfy enterprise procurement and security review.
The scope covered architecture, AI/ML, frontend, backend, dashboards, DevOps, and security. Compressing the full build into nine months required parallel workstreams across a cross-functional team with strong quality gates.
Solution
Solutions We Delivered
FiftyFive designed and delivered the platform as a production-grade SaaS marketplace built on microservices, not a scaled-up prototype. The architecture separates transactional integrity, search, and real-time state across purpose-fit data stores, so AI search and live availability scale independently of core booking flows. A domain-tuned NLP pipeline sits at the centre of the buyer experience, translating natural-language demand into ranked, deduplicated provider results. Security, GDPR compliance, and observability were engineered into the foundation from the first sprint. The full build shipped in 8.5 months across 2-week Agile sprints.
Architecture and infrastructure
FiftyFive built the platform as microservices on AWS, isolating each domain so search, booking, and provider services scale independently under load.
Microservices deployed on AWS ECS with AWS Lambda for event-driven workloads.
Node.js with TypeScript backend exposing GraphQL APIs for flexible, typed client queries.
Next.js and Tailwind CSS frontend for fast, server-rendered buyer and provider experiences.
Amazon S3 and CloudFront for asset storage and global content delivery.
Purpose-built database layer
FiftyFive separated the data layer by workload type so transactional integrity, search performance, and real-time state never compete for the same resources.
PostgreSQL as the system of record for bookings, accounts, and transactional data.
Elasticsearch powering full-text and AI-ranked search across provider catalogues.
Redis handling real-time caching for live availability and high-frequency reads.
AI and machine learning search engine
FiftyFive built a domain-specific NLP pipeline so buyers can describe training needs in plain business language and receive accurately matched results.
BERT-based embeddings tuned to corporate training terminology and buyer intent.
Domain-specific NLP pipeline for query interpretation and semantic matching.
ML-based ranking model integrated directly with Elasticsearch scoring.
Metadata-driven deduplication so near-identical provider listings resolve to a single result.
Core marketplace features
FiftyFive delivered both sides of the marketplace as complete products, giving buyers intelligent discovery and providers commercial visibility into their own performance.
Buyer experience combining AI natural-language search with structured filtered search.
Secure booking and in-platform messaging between buyers and providers.
Provider dashboards with catalogue management and performance analytics.
Multi-region and multi-currency support for UK, EU, and Middle East operations.
GDPR-compliant data handling and policy framework across user flows.
DevOps, security, and observability
FiftyFive engineered enterprise security and continuous delivery into the platform from the first sprint, so the build could pass enterprise procurement review without rework.
CI/CD pipelines using GitHub Actions and AWS CodePipeline for automated deployment.
JWT and AWS Cognito for authentication and role-based authorisation.
AWS KMS encryption protecting data at rest.
Amazon CloudWatch and ELK stack for monitoring, logging, and observability.
Test-driven development with Jest and Cypress enforcing quality gates throughout.
FiftyFive delivered the full SaaS marketplace in 8.5 months, ahead of the nine-month target. The platform is benchmarked for enterprise-scale traffic and has driven rapid provider onboarding across three regions.
100,000+ Concurrent Users
The platform is benchmarked to handle 100,000+ concurrent users with sub-200ms response times, supporting enterprise-scale marketplace traffic without degradation.
78% AI Search Adoption
78% of buyers chose AI-driven natural language search over structured queries, validating the domain-specific NLP engine against real user behaviour.
1,200+ Providers Onboarded
Over 1,200 training providers onboarded within the first six months, including many with no prior visibility in mainstream search engines.
Three-Region Expansion
The client launched the world's first free-to-use corporate training marketplace, expanding across UK, EU, and Middle East markets post-delivery.
An AI-powered training marketplace is a platform that connects corporate training buyers with providers using natural language search and machine learning ranking. Instead of browsing rigid course categories, buyers describe their training need in plain language and the system matches them to relevant providers automatically.
AI improves marketplace search by interpreting buyer intent rather than matching keywords. Domain-tuned NLP models convert free-form queries into semantic embeddings, then machine learning ranking scores results by relevance. On FiftyFive's corporate training platform, 78% of buyers chose AI search over structured filters.
A scalable SaaS marketplace typically combines a microservices backend, a purpose-split data layer, and a modern rendered frontend. FiftyFive built this training marketplace on Node.js with GraphQL, Next.js, PostgreSQL, Elasticsearch, and Redis, deployed as microservices on AWS ECS and Lambda.
A full production-grade SaaS marketplace build typically takes six to twelve months depending on scope, AI complexity, and compliance requirements. FiftyFive delivered a complete AI-powered corporate training marketplace — architecture, AI/ML, dashboards, DevOps, and security — in 8.5 months with a 12-member cross-functional team.
Scaling to 100,000+ concurrent users requires separating workloads across purpose-fit infrastructure. FiftyFive used microservices on AWS ECS and Lambda, PostgreSQL for transactions, Elasticsearch for search, and Redis for real-time caching, achieving benchmarked sub-200ms response times under enterprise-scale load.
GDPR compliance for an EdTech SaaS platform requires lawful data handling, consent management, defined retention policies, encryption, and auditable access control. FiftyFive built GDPR-compliant policies into this marketplace from the first sprint, using AWS Cognito for identity and AWS KMS encryption for data at rest.
BERT generates contextual embeddings that capture the meaning of a query rather than its literal words. In this training marketplace, FiftyFive tuned BERT-based embeddings to corporate training terminology and integrated ML ranking with Elasticsearch, so business-language queries return accurately matched provider courses.
Duplicate listings are handled through metadata normalisation and deduplication logic applied before ranking. Provider catalogues arrive in inconsistent formats, so the platform standardises course metadata, identifies near-identical entries, and resolves them to a single ranked result, keeping search output clean and trustworthy.
Microservices split a marketplace into independently deployable services, so search, booking, and provider management scale separately. A monolith scales as one unit, meaning a traffic spike in search consumes capacity needed for bookings. FiftyFive chose microservices on AWS for independent scaling and faster release cycles.
Two-sided marketplaces need supply and demand activated together. This platform onboarded over 1,200 training providers within six months by giving each side a purpose-built experience — AI-driven discovery for buyers, and dashboards with performance analytics for providers who previously had no search visibility.
Yes. Multi-region and multi-currency support requires regional data handling, currency conversion, and localised commercial logic built into the core platform rather than added later. FiftyFive delivered this training marketplace with multi-region and multi-currency support across UK, EU, and Middle East markets.
Building an AI SaaS product needs a cross-functional team spanning backend, frontend, AI/ML, UI/UX, and QA. FiftyFive delivered this platform with a 12-member team in two-week Agile sprints, offering a two-week free proof of concept and monthly billing on actual man-hours.
We use cookies to personalise content and ads, to provide social media
features and to analyse our traffic. We also share information about
your use of our site with our social media, advertising and analytics
partners who may combine it with other information that you’ve provided
to them or that they’ve collected from your use of their services.
Necessary cookies help make the website usable. Analytics cookies help
us understand how visitors interact with the website. Marketing
cookies may be used to deliver relevant ads and measure campaign
performance.
Necessary cookies
Required for core website functionality. These cannot be
disabled.
Analytics cookies
Helps us measure traffic and improve website performance.
Marketing cookies
Used for advertising, remarketing, and campaign tracking.
