Overview
Proposal · nothing is built yet
Shiffty is a proposed service where AI agents turn a store’s employee list, availability and traffic into a shift schedule, and then handle employees’ change requests in conversation.
The manager should describe the store once and review the result, not rebuild the schedule by hand every week.
Where the idea comes from
We used to build schedules for stores that ran sliding shifts: start and end times staggered through the day to follow customer traffic. It was a hassle, even after the training we gave the people who had to do it. Tools such as Deputy and the others in the competitive landscape show that software can take over much of this work. The question behind Shiffty is whether AI agents can make it simpler still, especially the back-and-forth with employees after the schedule is published.
What it would do
- Build a schedule. The manager provides employees, their availability and constraints, and the store’s traffic by hour. The agent proposes a schedule that covers the busy periods without breaking the rules. See Schedule generation.
- Handle changes through conversation. An employee asks for a day off, a swap or a later start. The agent checks what the change would break, suggests options, and sends a ready-made change to the manager. See Employee assistant.
Who it is for
Shiffty is for retail and only retail: supermarkets, grocery and chain stores whose hourly staff work the sales floor, the checkouts and the counters. Restaurants, warehouses, hospitals and other shift-based businesses are out of scope on purpose, so the rules, the traffic model and the language of the product can fit a store exactly.
The first reader we have in mind is a store or small chain manager who schedules hourly staff and whose traffic varies through the day and the week. Larger chains with their own workforce systems are not the first target. Which segment to start with is still open.
What is decided so far
The main ones: Shiffty is for retail only. Managers and employees will use one application, which reaches phones through Capacitor for now. And it will be built with the same tools as our other products. Everything else — including whether a store would choose it — is an open question.
What this site is not
This site describes an idea and the decisions around it. There is no product, no pilot and no customer yet. Every page says so at the top. Examples use made-up stores and people.
Where to go next
- The problem: why sliding-shift scheduling is hard to do by hand.
- User flow: what the manager and the employee would do, step by step.
- Architecture: one application for both, and what stands behind it.
- Market: who it is for, what already exists and where we think the gap is.
- DevOps: what is deployed today, which is only this site.
- Open questions: what we have to decide before building anything.
- Two weeks to a demo: the web-only plan for 18 October; Capacitor comes later.