Local-first computing is not nostalgia, it is a structural response to centralisation, fragility, and the erosion of digital autonomy. Systems like SchmidtPress, NeuralWeave, and my solar dashboards show why control, resilience, and ownership must return to the edge.
Why Local-first Computing Matters More Than Ever
Local-first computing is not a romantic longing for the early internet, nor is it a rejection of progress. It is a deliberate and increasingly necessary response to the consolidation of digital power, the fragility of cloud infrastructures, and the silent loss of user autonomy. Systems that run at the edge — SchmidtPress, NeuralWeave, my solar dashboards, or even a WordPress plugin that performs all logic locally — form a crucial counterweight to the centralising logic of modern cloud platforms.
For two decades the technology industry has promoted a single narrative: move everything to the cloud. This shift was sold as modernisation, efficiency, and inevitability, as though computing at the edge were merely a transitional phase between the personal computer and the corporate server farm. But the more we rely on cloud platforms, the more apparent their limitations become. Outsourcing is not neutral. It centralises control, creates structural dependencies, and erodes both resilience and digital sovereignty.
Local-first computing is not anti-cloud. It is anti-fragility. It challenges the assumption that infrastructure must live elsewhere and proposes an alternative: systems should work reliably, transparently, and autonomously on the devices we actually own.
SchmidtPress, A Case Study in Controlled Simplicity
SchmidtPress emerged from a simple need: a writing system that respected the act of writing rather than complicating it. Modern publishing platforms tend to accumulate dependencies, plugins, backend layers, and operational burdens that provide little benefit to the author. SchmidtPress avoids all of this by embracing a deliberately minimal architecture. Markdown files, a small index, and a lightweight client-side engine form a complete and independent publishing environment.
In more academic terms, SchmidtPress embodies architectural locality: the core logic, state, and data remain entirely under the author’s control. It demonstrates that meaningful publishing does not require a distributed backend or a multi-tier stack, only clarity of purpose and locality of execution.
Solar Dashboards, When Autonomy Produces Resilience
The same principle informs my solar dashboards. All essential computation — inverter readings, GTI calculations, AI-based forecasting, and visualisation — runs locally. The sole external dependency is a weather provider, and even that input is immediately stored and processed on my own systems. No cloud platform controls the workflow, the analysis, or the long-term access to historical data. Resilience here is not achieved by relying on redundant cloud regions, but through independence at the source: the system continues to operate regardless of what happens elsewhere.
This illustrates an important point: systems that live close to their data sources avoid latency, avoid dependency cascades, and maintain functionality even in the face of outages or policy changes. In many cases, this local approach outperforms cloud-based alternatives precisely because it avoids the structural fragility inherent in centralised services.
NeuralWeave AI, A Rebuttal to Cloud AI Orthodoxy
This becomes even more relevant in the domain of AI. Industry orthodoxy claims that serious machine learning workloads must run in the cloud. NeuralWeave offers a direct counterexample. All embeddings, indexes, inference pipelines, and search operations run locally, without handing data or control to external providers. Capabilities that cloud vendors package as “enterprise AI” can, in practice, be executed efficiently on local hardware without sacrificing performance, security, or privacy.
From a more academic perspective, NeuralWeave serves as a critique of computational centralisation. It demonstrates that semantic search, vector databases, and machine learning inference do not inherently require proprietary cloud infrastructures. They can exist at individual scale, under personal ownership, without performance penalties.
Even a WordPress Plugin Can Be Local-First
The same logic guided the redesign of my French Republican Calendar Converter. Instead of exporting functionality to a cloud API — the economically fashionable route — the plugin performs all computations in the browser. It operates as a fully self-contained tool, independent of hosting environments, service tiers, provider availability, or external endpoints. This shows that local-first design is not constrained by problem domain: whether converting historical dates, rendering Markdown, forecasting solar energy, or processing images, computation can occur directly where the user is.
Local-First as a Coherent Design Philosophy
Taken together — the dashboards, NeuralWeave, SchmidtPress, and the converter — these systems reveal a consistent philosophy. Local-first design is not a niche preference. It is a deliberate response to cloud fragility, centralisation, and the economic asymmetries that follow. Systems anchored at the edge preserve autonomy, maintain resilience, and resist the gravitational pull of platforms that aim to absorb everything into their orbit.
Cloud Fragility and the Illusion of Reliability
The fragility of cloud infrastructures is no longer theoretical. Outages from major providers regularly disrupt global services. Each incident reveals the risks of consolidation: a single configuration error, a failed update, or an oversized file in the wrong subsystem can cascade into worldwide downtime. The cloud is often described as resilient, yet its failure modes are large-scale and immediate.
Local-first systems fail differently. They degrade gracefully. They do not collapse because a remote service hiccups. They act as buffers against the systemic weaknesses of centralised platforms.
In academic language, we might say that local-first systems exhibit failure locality — the principle that failures remain contained rather than globally propagated.
A Political and Economic Dimension
Local-first computing carries an economic implication that is often ignored. Cloud infrastructure concentrates wealth in multinational corporations and extracts value from local ecosystems. Local-first tools, by contrast, direct resources toward local technicians, local hardware, and local expertise. The distinction is not merely technical. It is structural.
Choosing to build or maintain systems locally becomes a form of digital agency. It resists the consolidation of power in cloud oligopolies and strengthens decentralised, community-based infrastructures.
The Future Is Hybrid, But the Core Should Remain Local
Cloud platforms have their place. They are excellent for distribution, backup, and large-scale connectivity. But they should not be the foundation of everyday infrastructure. The core — the essential, the personal, the irreplaceable — should remain local.
SchmidtPress, NeuralWeave, my forecasting systems, and my plugin architecture are not rejections of modern computing, but examples of what computing looks like when we reclaim ownership of our tools. Local-first systems demonstrate that autonomy, efficiency, and resilience flourish best at the edge.
In a landscape increasingly defined by dependency, complexity, and centralisation, choosing local-first is not only technically sound, it is intellectually and politically necessary.
Why This Text
Part of the reason I care so much about local-first computing has nothing to do with ideology and everything to do with people. Several months ago, a friend of mine — a talented infrastructure engineer — was laid off. Not because he lacked skill or failed to deliver, but because his company’s customers demanded to move “everything that possibly can” into the cloud. Entire departments were labelled obsolete, not because their work vanished, but because a marketing narrative convinced management that owning infrastructure was somehow outdated.
He eventually found new work. But instead of the engineering role he had spent years mastering, he now works inside a massive datacenter operated by a well-known cloud provider — doing work that barely resembles engineering at all. His days consist of walking endless rows of server racks, scanning barcodes, replacing faulty machines, and updating tickets. Half the salary, half the responsibility, none of the creative or analytical work he once enjoyed.
This is the hidden cost of cloud centralisation. Companies do not merely outsource infrastructure. They outsource expertise. They hollow out their own technical teams, reduce engineering roles to vendor coordination, and transform once-specialised professions into repetitive, tightly supervised manual labour. The cloud does not eliminate work, it relocates it into environments where autonomy and craftsmanship matter far less.
Local-first systems resist that trend. They preserve skills. They keep knowledge alive. They give engineers something meaningful to build and maintain rather than something to merely service. And they remind us that computing does not have to devolve into an assembly line inside someone else’s warehouse.
This text, and the systems I build, are my way of resisting that loss. A reminder that we can still choose tools that empower rather than diminish us, and that technological progress does not have to come at the expense of human expertise.