Work

Concept builds with room to grow.

These concept builds show how Krythic approaches intelligent, practical software. As client engagements complete, published case studies will appear here — with permission, never invented.

Project concepts

Concept build

Dispatch board for a field service crew

A week-at-a-glance scheduling board meant to replace the whiteboard and the group chat: dispatchers see every job and every crew, technicians see only today.

Project details

Built to test one idea — that most scheduling software fails because it models the calendar rather than the work. A job carries its own state (quoted, scheduled, en route, blocked, done) and the board is a view of that state rather than the source of it, so a change made on a phone in a van and a change made at a desk cannot disagree. Drag-to-reschedule is matched by a keyboard command and a menu action, because a dispatcher on a trackpad and a technician on a cracked phone screen are not the same user. Runs on generated fixtures; no client data.

  • Custom software
  • Scheduling
  • Offline-tolerant
  • WCAG 2.2 AA

Concept build

Document intake with a human in the loop

An intake queue that reads incoming paperwork, proposes a classification and the fields it extracted, and asks a person to confirm rather than retype.

Project details

A study in where AI belongs in an operational workflow: the extraction and the ranking are automated, the decision is not. Every proposed value shows the passage it was read from, so a reviewer can check the source at a glance instead of opening the original document. Confidence is stated in words as well as colour, the least certain fields are focused first, and every accepted or corrected value is recorded with who changed it and when — which is what makes the automation auditable rather than merely fast. Built on synthetic documents; no client data.

  • AI integration
  • Human review
  • Audit trail
  • Document processing

What they have in common

A pattern, not a portfolio of screenshots.

Every build follows the same discipline: model the information clearly, design around the real workflow, automate the repeatable parts, and keep a human in control of the judgment calls. The domain changes; the standard doesn't.

Want to be the first case study?

Start a project