Work

Three bodies of work: 2 years inside Tekion, a solo product studio, and a masters research project taken to an investor deck. Each case study is a full page with its own sections; the machine copy of any of them is that URL with .md on the end.

Tekion Corp

Enterprise dealership software. Two years, 255+ tickets, 2 platforms.

• -Tekion Corp, overview The studio card for the whole run.
• -Tekion Corp, hub
• -Operations [Case Study] Parts Receiving Revamp.
• -Retail [Track] The dealership-facing side.

Dee Product

A solo product studio. Each app is designed, built and shipped end to end.

• -Dee Product, hub
• -Reelic [Shipped] Search every reel you ever saved.
• -Copy Cat [Shipped] A visual clipboard manager for Mac.
• -Nextdoor [Shipped] A private community app for one building, block or street.

Auxtail

• -Auxtail [Research] Commercialisation strategy for a patented towbar bike rack, from an MSc dissertation with Auxtail as the industrial partner.

---

The work itself

The short version of each case, in its own words.

Parts Receiving Revamp (Tekion Operations)

The problem was never the parts. It was the screen they arrive on: receiving was cluttered and complex, it was not integrated with the purchase order that raised it, and nothing could be actioned in bulk, so a clerk repeated the same steps line by line and left the module entirely to finish anything the purchase order owned.

My solution was 5 re-defined objectives, drawn out of a service blueprint, a heuristic evaluation, support tickets and a competitor matrix, then built as 16 web receiving screens with an OCR concept alongside. The output answers all 3 complaints and adds what the brief had not asked for: a keyboard friendly model, receiving on the web as well as on mobile, and a supplier's invoice read rather than typed. 5 dealerships were approached for moderated testing and 3 took part. The outcome has nothing to read against it yet: this never became a live product. The deck ends with phase 1 still to hand over, and its 3 success metrics were live user sentiment, support tickets and TAP requests.

Tekion Retail

A dealership runs its whole day on one piece of software. Retail is the half of it the dealership itself touches: the deal a salesperson builds at a desk, the money the cashier takes at the counter, the paperwork the customer leaves holding.

Four of these 5 shipped for Inchcape and Renault Retail Group on ARC; the fifth is a 3D pack that went out across the whole platform. Each began as a flow nobody had written down, and ended as a system you configure rather than rebuild for the next market.

Cashiering at Invoice Level. The problem was the unit. A dealership was paid one amount against one deal total, which is not how a customer actually pays: one invoice now, another later, part in cash and part on a card. The solution made the invoice the unit, each one payable on its own, in part, by any method, with all 23 payment paths drawn in 3 groups before a screen was designed. The outcome, as reported on the project's own brief, is 97% of cashiering use cases covered, satisfaction up 23% and a 93% overall score. The output is a framework running at 15+ UK dealerships and built to take the next market's rules.

Order Forms & Invoice. The problem was that every market wanted its own paperwork. A country's finance rules and its legal requirements each landed as another order form to build, and another form meant a developer, a release and a wait. The solution was a single template a dealership's own finance manager configures: any field renamed, given a sample value or switched off, against a live preview of the document the customer will receive. The outcome is that a new market is set up rather than rebuilt, by an admin rather than by engineering. The output is one order form and invoice system, built for Bravo Auto and integrated across the company and its clients.

Contract Management Module. The problem is written on the brief itself. Provision normally sits with the bank that owns the leased vehicle. At Renault Retail Group the bank and the dealership are subsidiaries of the same parent, so regulation puts the provision on the dealership instead. The solution is a module the dealership manages provision from, its 85 fields each given a source, RCI or ARC, and its error reporting drawn 6 ways before one was chosen. The outcome is provision held and tracked where the regulation puts it, by the party that answers for it. The output names its own next step: lease activity, active and inactive users, equity mining.

Activity & Audit Logs 2.0. The problem was that a log said something had happened without saying where. Reading one meant rebuilding the record it belonged to from memory, and the old system was deep enough that owning it took calls with senior designers, managers, product managers and developers before a line could be drawn. The solution, built with a senior designer, borrows a shape every user already has: the folder path from a desktop file browser, run across the top of the panel, with every entry coloured by the kind of change it records. The outcome is a log read at a glance, across 4 record types and 10 use cases. The output went back into the design system as new components and a new process for the teams after this one.

3D Placeholder. The problem was an empty slot. Every list, card and deal screen keeps room for a picture of the car, and with no photograph on file that room printed text. The design system had no standard thing to put there. The solution was to model the car nobody sells and render it from 6 angles, one for every slot the platform has. The outcome is a placeholder used across the whole platform, for GM, Ford, Aurora, Honda, RRG and Inchcape among others. The output is the pack itself, and the internal plugin built with a staff designer that renders the model live into the frame that needs it.

Reelic

Problem. Saving is not the broken part. That is one tap, and it always works. Looking is. A few thousand saves sit in one folder with no search box over it, so the only way back to something is to scroll until you see it again, carrying whatever you happen to remember.

The problem was never saving. That is one tap and it always works. It is looking: a few thousand saves sit in a folder with no search box over it.

My solution was to index them in your own browser, over 4 decisions, 3 of them harder on purpose. The output is a folder that answers from whatever you happen to remember, on Instagram, in a side panel or in a dashboard, live at 1.0.5 across 6 browsers. The outcome is small so far: weeks live and about 10 people using it, because I built the product before I built any way to find it.

Copy Cat

Problem. Remembering is not the broken part. Every clipboard manager already does that. Recognising is. What comes back is a list of grey rows that all look the same, so finding the one you copied earlier means reading the list line by line instead of seeing it.

The problem was never remembering. Every clipboard manager does that. It is recognising: what comes back is a list of grey rows that all look the same.

My solution was to make every copy a card, over 4 decisions that started with not writing the engine. The output is a strip you read by sight, each card carrying the app it came from, what it is and a preview of the thing itself, live at 1.0.0 with everything staying on your Mac. The outcome is that people pay for it, bought once rather than rented after 7 days free.

Next Door

Problem. Distance is not the broken part. A building already puts everyone within a floor of each other. The apps built for streets turned them into public feeds with ads in them, so the real conversation moved to a group chat, which nobody can search and everybody mutes.

The problem was never distance. The apps built for streets turned them into public feeds with ads in them, so the real conversation moved to a group chat nobody can search and everybody mutes.

My solution was one calm place for one building, over 4 decisions that all protect the same promise. The output is 3 surfaces of a single web app, chat, events and a marketplace, sharing one member list and one login, running inside free tiers with the code there to fork. The outcome has not happened yet: no street is living on it, because it is a demo rather than a launch.

Auxtail

The problem was never the engineering. Craig Broadbent spent 20 years designing towbars at Nissan and the patent was granted. Nobody had asked whether anyone would pay for it, at what price, or through whom.

My solution was to stop selling a rack and sell a system, priced off the value delta rather than the shelf, over 4 decisions that each name what being right costs. The output is a phased commercialisation strategy and the investor deck that carries it, from an MSc dissertation with Auxtail as the industrial partner. The outcome is not in yet: this is a plan, not a product in driveways.