← All guides
Local-first software

Local-first software: keeping custody when the cloud disappears

Years of land-trust maintenance logs, sensor annotations, and field photos accumulate inside a polished SaaS workspace. Then the vendor gets acquired, sunsetted, or prices out small stewards—and overnight the desktop client stops booting, leaving an engineer scrubbing through a mangled JSON or PDF dump at the eleventh hour. When the only canonical database lives on someone else's server, losing the subscription means losing operational memory.

Author: Escherbridge. Published . Last reviewed: , America/Denver.

That failure mode is the exact starting point of Martin Kleppmann, Adam Wiggins, Peter van Hardenberg, and Mark McGranaghan's April 2019 Ink & Switch essay, "Local-first software: You own your data, in spite of the cloud." As they note, cloud tools make collaboration fast—yet "by centralizing data storage on servers, cloud apps also take away ownership and agency from users. If a service shuts down, the software stops functioning, and data created with that software is lost." Their answer starts with one architectural inversion: treat "the copy of the data on your local device — your laptop, tablet, or phone — as the primary copy," while servers "hold secondary copies of your data in order to assist with access from multiple devices." Its first of seven ideals says it in five words: "1. No spinners: your work at your fingertips."

We dig into what that custody boundary means for field infrastructure in Local-first digital twin and A digital twin shouldn't live in someone else's cloud.

Real trade-offs behind treating the device as primary

Local-first is not an anti-server ideology—remote nodes are still vital for backups, relays, and multi-party sync. What changes is failure behavior: reads and writes succeed against local storage even on a flaky rural link, and when a sync network disappears temporarily, idempotent protocol retries and deterministic conflict handling keep state converging without holding the operator hostage. The real cost moves into engineering discipline: schema migrations outlive central control, key custody stays with the member (explored for browser passkeys in NEOS Keys in the browser), and conflict resolution rules must be explicit rather than hidden behind "last write wins."

Where custody connects across the architecture

Whether an organization stores facility state as a digital twin or coordinates field crews offline, the first architectural question is identical: if the host goes offline permanently tomorrow, does the operator still hold a runnable, verifiable copy?

Scope and boundaries

This note is grounded directly in the Ink & Switch essay rather than claiming CRDTs are deployed across every stack surface or inventing synthetic sync-conflict rates. PrimusNeo, NEOS, and AZOA are distinct infrastructure layers—this guide does not claim any of them as an off-the-shelf local-first consumer product.

A situation like this

A regional land stewardship team runs three years of site surveys in a proprietary hosted app. When the startup winds down on 60 days' notice, exported files break internal links because local custody was never part of the data model.

Sources

  1. Kleppmann, Martin, Adam Wiggins, Peter van Hardenberg, and Mark McGranaghan. "Local-first software: You own your data, in spite of the cloud." Ink & Switch, April 2019. https://www.inkandswitch.com/essay/local-first/

Also in this section

The useful next step is to read the 2019 essay, or to say where the primary copy of a tool actually lives. If something on this page is wrong, write to us.