Internal Enterprise CRM

Axact built and licensed enterprise software, with a catalogue running across ERP, CRM, HRM, recruitment, finance, payroll, procurement and user portals. Alongside the products it sold, it ran an internal department that built software for the company's own use. More than 2,000 employees worked inside those systems every day, which meant the internal tools carried a real operational load rather than sitting as side projects.

The CRM was one of those systems. It sat with the rest of the internal suite and served staff whose working day depended on it, which set the standard it had to meet: not a demo, not a prototype, but something people would be inside for eight hours.

There was no design. Not a thin design, not a late design: none. No comps, no wireframes, no specification, nobody upstream producing an interface for the CRM to be built from. The department had software developers and it had one person responsible for what the software looked like and how it behaved, and no process connecting the two.

The conventional answer in 2010 would have been to draw the screens in Photoshop, slice them, and hand the pieces to a developer to reassemble in markup. That route has a well-known cost even when a team is staffed for it: the drawing is not the thing, so every gap between the picture and the working page becomes a negotiation, and the person who drew it is not the person discovering the gaps.

Design in the browser, and be the person who ships it.

Rather than producing an artefact for someone else to translate, the interface was designed by writing it: layout, interaction and behaviour built directly in HTML, CSS and jQuery, in place, next to the developers building what sat behind it. The design decision and the implementation were a single act, so there was no translation step to lose anything in and no specification to keep synchronised with a running page.

Building the interface rather than drawing it was treated at the time as a constraint, the thing you did because there was no design process to lean on. It was actually the better method, and it took another fifteen years for the industry to arrive at the same conclusion and give it a name. The shortest distance between a design decision and a working product is to have the same person make both, and every argument for that in 2025 was already true in 2010.

The second thing is about the absence of tooling. No design system, no component library, no handoff tool, and the work still shipped and still held up under daily use by a large internal user base. The tools are an accelerant, not the capability. What carried the project was knowing what the screens needed to be and being able to make them exist.