Development
Delivery Models
UI/UX Design, Product Discovery, Business Analysis, Custom Web and Mobile App Development, Architecture Design and Cloud Infrastructure Setup, CI/CD and DevOps Automation, QA and Manual Testing, AI Agent Development, Post-MVP Support and Feature Scaling Strategy
Flutter, Node.js, ReactJS, AWS, AWS Bedrock
Twilio, GetStream, Google Maps, reCAPTCHA, WhatsApp Business API
GitHub CI, Flake8, Pytest, Pre-commit
The UK's railway sector had long relied on informal networks to get things done. Job opportunities spread through WhatsApp threads and industry groups — a system built on trust and personal relationships, but one that made it harder for skilled workers outside those networks to break in.
The hiring process itself was fast-moving and demanding. Contractors regularly received calls at any hour — and missing one could mean missing a contract.
Under UK regulation, contractors can work with up to three agencies simultaneously — one primary, two secondary — each managing its own compliance obligations and communication flows.
Connected Railway set out to change that — building a platform where contractors could find and apply for work, and agencies could source and manage candidates, without the constant back-and-forth of phone calls and informal channels.
We conducted in-depth research and multiple rounds of discovery calls with the people this platform was built to serve: contractors, recruiters, and training providers.
Openings lived in scattered chats — inconvenient to track, and easy to miss unless you were in the right group.
No centralised outreach or tracking. Matching ran on spreadsheets, phone calls and memory.
No standardised profile, no filters by training, location or availability — so last-minute hires and no-shows.
Contractors didn't know which certifications to pursue, or where a career could go next.
Without visibility into who was qualified and available, agencies defaulted to the contacts they already had.
In an industry built on precise scheduling and safety standards, there was no data to plan against.
Daniel wasn't looking for someone to simply build an app. He had no background in software and had never led a tech project. What he did have was a precise understanding of how UK railway hiring actually works — and a clear view of how technology could fix it.
We translated that knowledge into a product without jargon or pressure — mapping flows, testing features, and leaving room for feedback at every step. Daniel joined every call, challenged us to simplify, and kept the team focused on what contractors need.
We interviewed recruiters and contractors, ran market research and mapped feature expectations. Speed-to-market turned out to be decisive — so we defined a launch-ready scope with a clear post-launch roadmap instead of a full build.
The platform included four key components, each designed to serve a specific user group and deliver immediate operational value across mobile, web, and admin interfaces.
After the platform launched, the team focused on onboarding agencies and contractors and validating the core model. In spring 2026, the client returned with a new scope: three major feature areas that would significantly expand the platform's capabilities.
Phase 2 was delivered under Inoxoft's Product Development 2.0 model: two senior specialists, supported by custom AI agents, with the combined output of a full product team — business analysis, design, full-stack development, QA and DevOps, minus the handoffs and coordination overhead.
Turns business goals and raw requirements into development-ready specs. Identifies edge cases and flags gaps before a line of code is written.
Implements across the full stack from structured specs, cross-checks output against requirements, and loops until it's consistent.
Tests every output against spec, catches regressions and flags inconsistencies before human review. Quality gates sit inside the loop, not at the end.
Checks new screens and components against the established design system, keeping every surface coherent.
The AI agents running alongside the team are not generic tools. They were built and tuned by Inoxoft's R&D team, each one trained on real project experience — encoding Inoxoft's engineering standards, QA practices, design principles and delivery methodology accumulated across dozens of shipped products. They carry the institutional knowledge of a senior team.
The scope for Phase 2 was fixed and agreed upfront. Working within that scope, the team delivered 2–4× the output of a comparably sized conventional team — covering, across two engineers and custom AI agents, every role that Product Development 2.0 accounts for: product strategy, business analysis, design, full-stack development, QA and DevOps.
With the Development Agent handling syntax, patterns and platform-specific implementation, engineers are no longer constrained by their primary stack. A mobile specialist can confidently ship cross-platform and web code — the agent carries the technical context, the engineer provides the product judgment. In practice, the team can cover virtually any language, platform or framework the project requires.
Updating registration flows, phone number formatting and Eircode validation across the platform — completed in a fraction of the time a conventional team would require.
The training providers module, WhatsApp integration, Irish market adaptation and all additional features were delivered through the Product Development 2.0 model — each output reviewed and approved by the engineers before reaching production.
By the time Phase 2 closed, the team had delivered 2–4× the output of a standard delivery model. Every improvement the platform needed was identified mid-flight, scoped in days and shipped in the same window — without adding headcount or compromising quality.
Generic AI makes each engineer quicker at the task in front of them. It doesn't remove the handoffs, the role gaps or the coordination overhead — so the scope stays the scope.
Our agents encode Inoxoft's own analysis, engineering, QA and design standards, so the loop absorbs whole roles. That's why Phase 2 came back both faster and bigger than it was scoped.
One account now posts both jobs and courses — a structural change to how the marketplace works.
Provider-specific fields that render consistently across web and mobile, with a small engineering footprint.
Agencies can publish before a rate is agreed, matching how negotiations actually happen.
A rebuilt agency workspace, reflecting how the platform is used at 40+ agencies rather than at launch.
Every enrollment tracked in one place, from first conversation through to completion.
Filtering by course name, company and location across the whole provider catalogue.
A dedicated contact page per provider, so contractors reach the right person directly.
Railway certifications and terminology explained in-product, for contractors new to the sector.
