Services / 05

$ cat /services/design-research

Design Research & Systems

Ethnography before electronics. The data layer after.

hours of field observation before design
tick-to-screen latency
minutes · a workflow, before and after
ETHNOGRAPHY → ELECTRONICS → DATA LAYER FIELD 50 HRS MAP JOURNEYS SPEC DEVICE+DATA BUILD REVISIONS PIPELINE <100 MS DASHBOARD LIVE WHAT THE DASHBOARD SHOWS, THE FIELD EXPLAINS
FIG S5 · FIELD TO DASHBOARD · ONE SIGNAL CHAIN

We spent fifty hours behind a bar before designing a pour sensor. Watching the work set the spec: how bottles are racked, what gets dropped, what goes through the glasswasher that was never supposed to. Devices designed from the field get used. Devices designed from a conference room get shelved.

The research runs through to the software that makes the hardware worth owning: real-time pipelines, dashboards, alerting, on-premise AI where the data cannot leave the building. A proprietary trading desk we rebuilt inside a spreadsheet holds tick-to-screen latency under 100 ms and cut the morning workflow from 35 minutes to 5.

Method

  1. Observe

    Hours in the field before hours at the bench. The workflow as it is, with its workarounds, is the true requirements document.

  2. Map

    Journeys, failure points, and teardowns of the tools people already trust. What survives contact gets kept.

  3. Spec the pair

    The device and its data contract are designed together, so the sensor never ships an orphan.

  4. Ship the system

    Pipeline, dashboard, and alerts delivered with the hardware as one artifact. A sensor without its data layer is a paperweight.

Deliverables

  • Field studies & usability testing
  • Journey maps & competitive teardowns
  • Device + data-contract specifications
  • Real-time pipelines & dashboards
  • On-premise AI integration
  • Alerting & reporting layers

Bench

  • Field study & shadowing kits the bar, the floor, the desk
  • Usability protocols tested on the people who will use it
  • Real-time dashboards WebSocket-fed, sub-second
  • On-premise AI when the data cannot leave the building

Proof