Why Local-First Is Ready to Challenge Your Cloud-Only Stack

Adam Wiggins, Heroku co-creator, argues the pendulum swings back from cloud-dependence to local-first architectures. Tools like Linear show CRDT-based sync engines deliver instant UI by storing data locally and syncing in the background. This isn’t about abandoning the cloud, but strategic placement of compute and data for performance and resilience.

The real signal for engineering leaders is that these patterns are now production-ready for certain domains—collaborative tools, productivity apps—but not for all (e.g., banking). The tradeoff is increased complexity in conflict resolution and domain modeling. Teams must evaluate if partial local-first adoption can reduce infrastructure costs and improve user experience under unreliable connectivity.

AI agents are accelerating this trend: heavy cloud dependency frustrates developers who want to work offline. Local models and sync engines could allow 80% of tasks to run locally, reserving cloud for complex queries. This mirrors the Git/GitHub model where local work and central collaboration coexist.

Leaders should watch for mature sync engines and identity/auth innovations. The Local First Conference community is bridging academic research with pragmatic engineering, signaling a shift worth tracking for long-term architecture decisions.