RicherTask
Task and weekly planning for the college's staff, in Hebrew, built around how the team already worked instead of a methodology they would have to adopt.
- Client
- Richer LTD
- Role
- Sole engineer
- Period
- 2025–2026
- Stack
- Next.js, TypeScript, Supabase, Row Level Security, Vercel
- 1,300+
- Commits
- Daily, in production
- Use
- Per-user RLS
- Access model
The problem
Work lived in WhatsApp messages and people's heads. Anything not urgent was invisible until it became urgent. Off-the-shelf tools had been tried and abandoned twice, because they asked the team to adopt a workflow rather than reflecting the one they had.
eitanrafael.dev/demo/richertaskopenWhat I built
- Modelled the team's real units of work — personal tasks, weekly planning, workspaces and projects — rather than imposing a sprint or kanban structure.
- Made the weekly view the default screen, because the team plans in weeks and an unfinished task rolling forward is the normal case, not an exception.
- Enforced access per user in the database with Row Level Security, so a query cannot return another person's tasks regardless of what the client asks for.
- Built it in Hebrew and right-to-left from the first commit, not as a translation layer added afterwards.
The hard part
Adoption, not features
Two tools had already failed here, so the risk was never technical. It was that staff would open it once and go back to WhatsApp. That constraint drove the design: the weekly view is the landing screen, rolling a task forward takes one action, and nothing requires a status update that exists only to feed a report.
Outcome
It became the tool the team actually opens, across 1,300+ commits of iteration driven by watching people use it rather than by a roadmap.