Apps, websites and software systems

Checklists for the working model (step 2) and the feasibility check (step 4) when the “thing” is software, or a device that talks to other devices.

The working model

Users and their jobs: who uses it, for what, how often, and on which devices.
Data: what it is, where it lives, who owns it, how big it gets and how fast it grows. Which copy counts as the true one?
Flows and state: what happens from the user’s action to the stored result. What states can each item be in?
Sync and offline: can two copies change at the same time? How do they merge (last-write-wins, merge rules, CRDTs)? What happens while a device is offline, and when it comes back?
Connectivity: how the devices actually reach each other (internet, local Wi-Fi, Bluetooth, cable, a file). Is that path available when it’s needed?
Integrations: other services and APIs, with their limits, costs and rules.
Accounts and security: login, permissions, what an attacker would target, and where secrets are kept.
Privacy: personal data, where it’s stored, and the laws that apply (GDPR and others).
Failure cases: network loss, partial failure, retries and duplicates, a crash mid-write, a full disk.
Scale and cost: users, requests per second, storage, with rough numbers. Hosting cost per month.
Platform limits:
- iOS and Android restrict background work, Bluetooth and local-network access, and require permissions.
- Browsers sandbox files, storage and devices.
- App stores have review rules.

Possible: does the physics or logic allow it? For example, two devices can’t sync without some path between them, and “real time” across the world has latency.
Buildable: by whom, with which skills, in what time?
Operable: who keeps it running, updates it and fixes it at 3 a.m.?
Legal: privacy, licences, store rules.

Build vs. buy vs. use an existing service or open-source tool. Name one existing option if it exists.
The simplest architecture that meets the need. Every server, queue and sync layer is something to run and debug.
Recommend one approach, and say what would make you pick a different one.