Six projects, and what actually changed.

The problem

Every quote was assembled by hand from carrier rate sheets in email. Turnaround averaged three days, and roughly one in six quotes went out with a pricing error the ops team had to absorb.

What we did

1. 1Mapped the full quoting path with the ops team before designing anything
2. 2Built a rate engine ingesting carrier sheets on a schedule, with versioned pricing
3. 3Shipped a customer portal for self-serve quotes, booking and shipment status
4. 4Added margin guardrails that flag any quote below threshold for approval

The problem

Four staff spent most of each morning transcribing orders from WhatsApp messages, photographed handwritten lists and voice notes in three languages. Mistakes surfaced only at delivery.

What we did

1. 1Built an ingestion layer for text, image and audio orders from one WhatsApp number
2. 2Extraction into a strict schema, matched against the live SKU catalogue
3. 3Low-confidence lines route to a human review queue instead of guessing
4. 4Nightly eval suite over a labelled order set catches accuracy drift

The problem

Six clinics shared one slow WordPress site. Mobile load time was over eleven seconds, the booking form broke below 400px, and no location page ranked for its own suburb.

What we did

1. 1Rebuilt on a static-first stack with per-location pages and local schema
2. 2Redesigned booking as a three-step mobile-first flow with no dead ends
3. 3Gave non-technical staff CMS control of hours, pricing and practitioners
4. 4Full WCAG 2.2 AA pass, including screen-reader testing of the booking flow

The problem

Coaches were sending programmes as PDFs and tracking results in spreadsheets. Members dropped off in week two, and the gym had no visibility into why.

What we did

1. 1One React Native codebase shipping to both stores from the same CI pipeline
2. 2Offline-first logging with conflict-safe sync for basement gyms with no signal
3. 3Coach web dashboard for programme building and adherence tracking
4. 4Instrumented the week-two drop-off so retention work had something to aim at

The problem

Analysts rebuilt the same quarterly deck for 90 clients in Excel and PowerPoint. The cycle took two weeks, and figures occasionally disagreed between documents.

What we did

1. 1Single data pipeline so every figure has one source and one definition
2. 2Role-scoped portal: clients see only their holdings, advisors see their book
3. 3Scheduled PDF generation preserving the firm's existing report design
4. 4Full audit log of who viewed and exported what, for compliance

The problem

The studio's work is visual and their old site flattened it — compressed thumbnails, no sense of space, and a 14MB homepage that lost most mobile visitors before it painted.

What we did

1. 1Art-directed project pages with scroll-paced reveals instead of a uniform grid
2. 2Responsive image pipeline with AVIF and per-breakpoint crops
3. 3Enquiry flow that captures budget and timeline before the first call
4. 4Reduced-motion variant that removes parallax without losing the layout