Platforms that have carried the load for years

Starting point

For a vehicle manufacturer, telematics stopped being an accessory long ago. It is the foundation of the business: fleet operators expect to see their vehicles in real time, to evidence cold chains without a single gap and to head off breakdowns before they happen. Our customer’s on-premise architecture had been built for a far smaller fleet and was visibly running out of road. Every newly connected vehicle added load, scaling meant procuring hardware rather than pressing a button, and the stability risk grew quarter by quarter. One thing was clear: the future of the business lies in data-driven services, and that calls for a foundation that carries today’s operation and opens the door to what comes next.

So the brief went well beyond a straightforward migration. The platform had to be rebuilt cloud-native, absorb the existing fleet without a minute of downtime and serve from day one as the digital core for new business models.

The solution

We rebuilt the platform from the ground up as a distributed, horizontally scalable cloud architecture and migrated the live fleet in just 14 months. Instead of copying the old architecture into the cloud (“lift and shift”), we rethought it: event streaming as the backbone for message processing, container orchestration for elastic scaling, purpose-built stores for analyzing large volumes of sensor data. We describe the entire infrastructure as code, so every environment is reproducible and every change auditable.

Today the platform processes telemetry from more than 150,000 actively connected vehicles, and it does so across manufacturers: it takes in our customer’s own vehicles alongside third-party makes, whatever telematics generation is fitted. Each vehicle reports more than 600 distinct sensor values, from position and temperature to tyre pressure, door status and braking data. In normal operation the system handles 250 to 350 messages per second and absorbs peaks of more than 1,000 messages per second without breaking stride.

On top of that we built a portal for fleet operators that goes far beyond a dot on a map: real-time visibility of the whole fleet whatever the hardware, freely configurable alerts and notifications, complete histories of events, positions and temperatures. At order level, operators track journeys live, with arrival forecasts and the temperature reports the cold chain demands generated automatically. A monitoring module pairs proactive surveillance with predictive maintenance and keeps a close eye on the refrigeration units: irregularities surface before they turn into failures, and maintenance is planned in advance rather than improvised once something breaks.

Communication runs both ways. The platform does not just receive data: it commands functions such as the refrigeration unit or door control remotely and pushes over-the-air updates down the same channel. A dedicated update management system plans and documents these campaigns in line with the regulatory requirements for in-vehicle software updates, so every update is auditable and every vehicle stays compliant without anyone having to touch it.

And the platform has become the commercial foundation it was always meant to be: we have put an AI-powered assistant into production that answers users’ questions about vehicles and their data in plain language, grounded in the live data context.

The result

Since going live in the cloud, the connected fleet has grown from around 30,000 vehicles to more than 150,000 — with no architectural rebuild, because the design scales past 600,000. The platform now runs the telematics operation of one of Europe’s largest connected trailer fleets. It is also anything but a finished project: under our Platform Stewardship we have secured, operated and continuously evolved the system ever since go-live, with the same team, monitored around the clock, with automated load and integration tests in every release.

Starting point

It sounds like a detail, yet it is a problem with many dimensions: every vehicle built needs the right combination of tyres, wheels and valves, right down to the specific wheel assembly. What looks trivial is in fact a high-dimensional optimization problem, because the criteria compete with one another: cost points to one combination, the efficiency ratings that drive CO₂ compliance to another, contractually guaranteed supply shares and supplier capacity ceilings to a third. Optimize for one and you inevitably breach another.

Until now the assignment was made largely by hand: a laborious step that tied up specialist capacity, ran for hours and left nobody able to prove the result was genuinely the best across every objective. Our customer wanted a solution that makes the call automatically and demonstrably well, and slots into the existing configuration process rather than forcing a redesign of it.

The solution

Classic rule-based automation breaks down on problems like this: there are far too many combinations to work through, and rigid rules cannot balance competing objectives. So we implemented an evolutionary optimization method, an approach borrowed from nature: the system scores an assignment, mutates it along defined rules and keeps only the changes that improve the overall result. Generation after generation it converges on a solution that performs strongly across every objective, rather than excelling on one criterion at the expense of all the rest.

The selection runs in stages: tyres first, then wheels and valves, finally the specific wheel assembly. Commercial, environmental, technical and contractual criteria all count at the same time. Each component can carry several weighted objectives, and the weightings are configurable, so shifting priorities no longer mean a development project. The system also spreads volumes across the time blocks as evenly as possible between suppliers, to avoid bunching in procurement.

We wanted the system to be honest with its users. Assignments it cannot resolve, and capacity ceilings it would have to exceed, are never quietly fudged: they come back through a structured alert, and the call on how to handle them stays with people. Simulation runs let the business try out alternative objectives and time-period splits before anything is scheduled for real: what does it cost to weight the efficiency targets more heavily? What happens if a supplier drops out? A BI dashboard puts the effect of every optimization run in plain sight, with figures such as average efficiency ratings and cost before and after the run.

The result

A manual process that ran for hours and tied up specialist capacity is now an automated, auditable run of under five minutes, mathematically optimized across 15,000 orders and more than 300 materials, with contractual supply shares and capacity limits honoured every time. And the gain is not a claim but a measurement: the dashboard reports it run by run.

Starting point

A global sales force has a structural problem: the knowledge of the very best salespeople sits in their heads and in scattered file shares. Which arguments land, which documents convince, which options suit which customer. Every region pitches differently, material goes out of date unnoticed, and regional specifics such as winter equipment in the north or sun protection in southern markets have to be hunted down every single time. Then there is the practical hurdle: sales conversations happen wherever the customer is, on depot forecourts, in workshops, in regions with no reliable network coverage.

Our customer wanted one tool for the whole world, letting sales staff assemble and present a tailored showcase for every customer: multilingual, regionally appropriate, centrally maintained and still ready to go anywhere, offline included.

The solution

We built a guided-selling application that puts the pooled expertise of the best salespeople within everyone’s reach. At its heart sits a flexible presenter that organizes the right material by category, from general product information through flyers, brochures and value arguments to photographs and video. Products, campaigns and their supporting documents are managed centrally and configured to the customer situation at hand: the salesperson assembles the pitch from approved, current building blocks instead of rebuilding it from scratch every time.

Underneath sits a central media asset management system: marketing images, graphics, documents and video are versioned, held in every language and found in seconds through bespoke keywords. An administration interface covers product management, the mapping of data and copy, the language variants and the roles and permissions. Maintenance stays central, use stays global.

The real engineering challenge was working offline. We designed the application local-first throughout: the content a user needs is checked out in advance and held on the device, and a local database keeps the application fully usable with no connection at all. A robust synchronization mechanism pushes subscribed content out to the clients and refreshes it the moment a connection returns. Users see none of that machinery; their material is simply there and up to date. A responsive design carries it from notebook to tablet, in the meeting room and on the depot forecourt alike.

The result

Sales teams worldwide gained one tool for addressing customers personally, in their own language and to regional expectations, online or offline and on any device. Polished, always-current sales material is available anywhere on the planet, and the knowledge of the strongest salespeople reaches every conversation, not just the ones led by the most experienced.

Starting point

Controlling who may access what in an industrial company with thousands of employees means fighting on several fronts at once: joiners, leavers and internal moves have to translate reliably into permissions; devices and software need ordering, assigning and managing; and IT costs should end up with the divisions that incur them. At our customer these processes ran through a scatter of ad-hoc channels: emails, corridor requests, paper forms. That created work and cost the business its audit trail. The critical part: proving that every last access has been revoked when someone leaves is effectively impossible through ad-hoc channels, and it is the first thing any auditor looks for.

What was needed was a single system to govern identities, permissions and the IT operating processes around them across the entire company, from an employee’s first working day to their last.

The solution

The system we built runs on- and offboarding end to end: on arrival, access, rights and equipment are provisioned reliably and on the record; on departure, every one of them is withdrawn. It sounds like table stakes, yet in organizations that have grown over decades it is rarely done cleanly, and it is one of the most important compliance building blocks there is: no orphaned account survives, and every permission carries a documented reason and a documented end.

Self-service functions are central to it, giving every employee a structured route through the everyday IT procedures: ordering software and devices, requesting access to systems, folders and mailboxes, each through configurable approval paths that bend to the organization rather than the other way round. Requests run in the open and on the record, routine work stops landing on the IT team, and the team spends its time where judgement is actually needed.

The system also manages the company’s assets, devices and software alike, and handles the internal recharging of IT services automatically, reporting included: every division sees which IT services it consumes, and cost allocation runs in the background without anyone maintaining a spreadsheet.

The result

Our customer now governs identities, permissions and IT operating processes from one place: less manual work, a clean audit trail, costs that land with the teams that incur them and an authorization model that stands up to scrutiny. Once the system was live we handed it over to the customer’s own development team in good order, documentation and knowledge transfer included, and they have run and extended it themselves ever since. That, too, is what partnership means to us: we do not lock customers in, we equip them.

Starting point

At the end of the nineties nobody said “IoT”, but the problem was already there: an energy utility runs thousands of industrial plant components spread across a large network, and the day-to-day operation of that infrastructure rested on a patchwork of siloed systems and decentralized shadow IT. Faults came to light once they had already taken effect, and operational processes such as maintenance, incident management and asset administration ran alongside one another instead of into one another. For infrastructure that heats entire regions, that was never going to hold.

Off-the-shelf software could not meet requirements this extreme: the sheer number and variety of assets, the depth of integration, the governance obligations that come with critical infrastructure. So the platform was built bespoke, from the ground up, alongside the customer’s own business units.

The solution

The platform manages thousands of heat and energy components with millions of sensors across the whole network. It processes telemetry and sensor data from generation and distribution in real time and keeps the condition of every asset visible, from a single unit to the network as a whole. On that data it detects anomalies, supports predictive maintenance and, when something looks wrong, sets automated incident management in motion: a reading becomes a fault report, the fault report becomes a dispatched job, with no manual re-keying in between.

The platform is far more than IoT. It runs the operational business end to end: project planning, maintenance and incident management, plant and network administration, meter management, right through to the handling of hazardous substances. Joining machine data to business processes is what turns readings into decisions and decisions into action.

Much of the work lay in integration. Hundreds of interfaces connect the platform to the existing systems and plant, including secure integration with control systems through established automation standards such as OPC UA. For governance reasons, and with the requirements for critical infrastructure (KRITIS) in view, the solution is deliberately hybrid: cloud services combined with operation in dedicated environments wherever data sovereignty and regulation demand it. A high-availability architecture with cross-site redundancy and comprehensive backup strategies keeps operation dependable throughout. For critical infrastructure that is not a nice-to-have, it is an obligation.

And then there is the dimension that sets this reference apart: time. For 24 years we operated the platform as Platform Stewards and kept evolving it, through several technology generations, shifting market conditions and regulatory demands that only ever grew. What began in 1999 was never “finished” and never turned into legacy: it grew with the business.

The result

Infrastructure that helps heat and power the lives of millions: real-time monitoring, predictive maintenance, operational process handling and end-to-end integration in a single solution, under the strict governance requirements of critical infrastructure. In 2024 we handed the platform over to the customer’s internal IT in good order, with complete documentation and a structured knowledge transfer. A quarter of a century of partnership that stayed professional right through to the goodbye: perhaps the strongest evidence of what Platform Stewardship means to us.

Starting point

In the international marine fuels trade, risk assessment is not administration, it is survival: every bunkering is a credit decision, every portfolio shifts by the minute, and counterparties are spread right across the globe. Our customer’s order and risk management system had grown up over years and was no longer equal to that business: risk was assessed in overnight batch runs, so anyone looking at the numbers in the morning was looking at yesterday. In a time-critical, globally distributed trade, that is not merely inconvenient, it is dangerous.

On top of that came the classic dilemma of the irreplaceable: the system was business-critical, woven deep into daily operations, and still could not stay as it was. Switching it off was impossible, running it on was risky. Precisely the situation we are built for.

The solution

We rebuilt the platform from scratch and turned the architectural principle on its head. Instead of collecting data and crunching it overnight, the new system is thoroughly event-driven: every new piece of information, whether an order, a payment or a limit change, immediately triggers exactly the calculations and process steps it affects. At the core sits a risk calculation engine that processes incoming financial data in real time and recalculates exposure on the spot. A change in the portfolio shows up in the risk figures within seconds instead of the following morning. On more than €1bn of annual risk volume, that means the assessment holds at any moment of the day, not once every 24 hours.

Beyond risk assessment, the platform covers order management end to end, from capture through to settlement. The decision processes around it are configurable: approval paths and rules adapt to the demands of different markets and trading situations without a line of code changing. Process expertise sits where it belongs, with the business, not buried in the source code.

The most delicate phase was the switchover itself. You cannot pause a system that live trading runs on. So we ran the migration as a controlled cutover in flight, prepared through parallel operation and comparison testing until the new system had proved itself. Trading never stopped, and a high-availability architecture with hot standby has kept operation unbroken ever since.

The result

More than 1,000 users in over 100 countries work on the platform today, and it processes more than 100,000 bunkering transactions a year. Overnight batch runs became risk figures in seconds, and a legacy system nobody wanted to touch became a modern, event-driven platform with room to grow. It also settled the question of whether the irreplaceable can be replaced — in flight, without pausing the business.