The challenge
Our client is one of the world's largest talent acquisition outsourcers, managing complex, high-volume recruitment programmes for enterprise clients across the globe. Their recruiters work across multiple platforms simultaneously — sourcing candidates from job boards and professional networks, managing communications, writing and optimising job content, analysing labour market data, and tracking hiring performance. Each of those activities typically lives in a different tool.
The goal was to change that: to build a cohesive suite of AI-powered recruiting tools that would sit across the client's existing workflows, reduce the time spent on repetitive high-effort tasks, and give recruiters genuinely useful intelligence at the right moment. The platform needed to meet enterprise security and compliance requirements — the client holds ISO 27001:2022 and ISO 42001 certifications — and deploy rapidly into client environments with minimal IT overhead.
The engineering challenge was not simply building capable tools. It was building tools that recruiters would actually choose to use, within the workflows they already had, without requiring them to change how they work.
The application
The suite is built as a family of Ruby on Rails applications, each addressing a distinct area of the recruiting workflow, unified around a central API layer that handles shared data, authentication, and integrations. Keeping the applications discrete allowed each product to evolve at its own pace while the API layer ensured consistent behaviour and data across the suite.
The AI assistant — generative AI for recruiters
The AI assistant is the core of the platform, integrating directly with the client's proprietary AI API to bring generative capabilities into everyday recruiting tasks. Recruiters use it to generate and optimise job descriptions, produce personalised candidate communications, create candidate summaries with gap analysis against role requirements, and work through custom prompts for tasks that don't fit a standard workflow.
The UX of the AI assistant was a significant focus throughout development. Generative AI tools are only useful if the output is trustworthy and the interface doesn't add friction to an already demanding workload. We designed the application around the recruiter's actual task context — the inputs are structured to guide the model towards relevant, accurate output, and the results are editable and actionable directly within the tool. The gap analysis feature, in particular, required careful design: presenting structured comparison data in a way that is genuinely useful for a hiring decision rather than a wall of generated text.
Candidate sourcing and screening
The sourcing application aggregates candidate sourcing across major job platforms and professional networks, pulling data into a unified view and applying automated screening and prioritisation. Integration with market-leading sourcing platforms means recruiters are not logging into individual systems; the platform brings the candidates to them.
The application includes automated interview scheduling and candidate summary generation, both of which required careful orchestration between the API layer, external calendar integrations, and the Rails front-end to maintain a fluid experience under conditions — high concurrency, multiple external API dependencies — where the user experience can easily degrade.
Insights — recruitment analytics
Insights provides the analytics layer: talent motivation data, competitive positioning, churn forecasting, and offer targeting intelligence drawn from integrated data sources. The application surfaces job insight data and market metrics in a form that supports actual decisions, rather than simply presenting numbers. The design work here centred on making complex labour market data legible to recruiters who are experts in hiring, not data analysis.
The browser extension
One of the most consequential UX decisions in the project was the Chrome and Edge browser extension. Recruiters spend their working day inside their existing tools — ATS platforms, LinkedIn, job boards — and asking them to context-switch to a separate application for every AI-assisted task creates enough friction to undermine adoption. The extension brings the full suite into the recruiter's existing environment, surfacing the right tool at the right moment without requiring them to leave the page they are working on.
The extension communicates with the central API layer, so every feature available in the web applications is available in context within the browser. It also means that as the platform grows, new capabilities are available in the extension without requiring a separate release cycle.
API-driven email delivery platform
Personalised, high-volume candidate communications require a delivery platform that is reliable, measurable, and capable of handling enterprise-scale sending volumes. We designed and built an API-driven email delivery platform built around AWS SES, giving the suite a robust, monitored sending infrastructure with delivery tracking and bounce handling built in. Candidate communications sent through the AI assistant and the sourcing application route through this platform, ensuring consistency of delivery behaviour across the suite.
The infrastructure
The platform is hosted on AWS in a configuration that reflects the availability and performance requirements of enterprise recruitment operations — where downtime or slow response times directly affect hiring programmes running to client SLAs.
The core Rails applications run on EC2 instances behind Application Load Balancers, providing horizontal scalability and automatic traffic distribution. The database tier runs on RDS with Multi-AZ failover. For services requiring containerised deployment and independent scaling, we use ECS Fargate — abstracting away instance management and allowing compute to scale to demand without manual intervention. AWS Lambda handles event-driven workloads: asynchronous processing, webhook handling, and tasks where a persistent server would be disproportionate to the work involved.
The result is an architecture that is appropriately sized for each type of workload: persistent and scalable for the application tier, managed and resilient for the data tier, elastic and cost-efficient for background and event-driven processing. SES sits within the same AWS environment as the rest of the platform, keeping email delivery latency low and operational monitoring consolidated.
The outcome
The platform is now in active use across the client's enterprise base, supporting recruiters throughout the hiring journey — from job content and candidate sourcing through to screening, communications, and analytics.
The multi-application architecture has proved its value as the suite has grown: new capabilities have been added to individual products without disturbing the others, and the central API layer has absorbed new integrations without requiring changes to the front-end applications. The browser extension has become a significant part of how recruiters interact with the platform day-to-day, validating the decision to bring the tools to where recruiters already are rather than asking them to come to a separate interface.
On the infrastructure side, the mix of EC2, Fargate, and Lambda means the platform meets enterprise SLA requirements without over-engineering every component. The platform is observable, resilient, and structured to grow with the client's requirements.