Writing · Nº 041065 essays
The writing
Ideas from inside the work. Explore technology, entrepreneurship, and the practice of building things that last.
29 essays found · Page 2 of 3
Clear filtersOperatingA Hybrid Stack: Self-Hosted Core, Cloud for OverflowA hybrid stack keeps your steady core self-hosted and rents cloud only for spiky overflow. How to split workloads so you own the base and pay only for peaks.EngineeringFall Back to Deterministic When the AI Is UnsureThe strongest AI-native systems keep a deterministic path underneath the model. When to route to rules instead of the model, and why a hybrid beats pure AI in production.StrategyServerless Is the Deepest Lock-In There IsServerless feels like freedom but is the deepest vendor lock-in there is, because your app gets welded to one provider's primitives and cannot leave without a rewrite.EngineeringThe Technical Debt Hiding in Bolted-On AIBolting AI onto old software creates technical debt that compounds fast. Where the debt hides, why it grows, and what it costs when you finally have to rebuild.StrategyMulti-Cloud Is Not the Same as No Lock-InMulti-cloud is sold as freedom from lock-in, but running on two clouds usually doubles your lock-in. What actually removes lock-in, and what just spreads it.StrategyWhen to Fork a Product Out of the Shared FoundationWhen should a product leave the shared foundation? When its needs diverge so far that the foundation costs more than it saves. Here are the signals to fork out.EngineeringHow to Design a Data Pipeline for Scraped DataHow to design a data pipeline for scraped data: separate collection from processing, stage raw HTML, validate before storage, and make reprocessing cheap.EngineeringWhy AI-Native Products Need a Different Data ModelAI-native products need a different data model than normal software. Why the schema has to store provenance, versions, and model reasoning, not just final values.EngineeringAPI Lock-In Is Worse Than Data Lock-InData lock-in gets all the attention, but API lock-in is worse. When your code is wired to a proprietary API, leaving means a rewrite, not an export.EngineeringHow to Design Pub/Sub Channel GranularityChannel granularity is the biggest pub/sub design decision. Too coarse and you leak data and waste bandwidth; too fine and you drown in subscriptions.EngineeringMistakes to Avoid When Building a Shared FoundationThe common mistakes that turn a shared foundation into dead weight: premature abstraction, no owner, leaky product logic, and versioning you skipped.EngineeringWhat Actually Changes in AI-Native ArchitectureAI-native architecture is not a chatbot on a normal app. Here is what actually changes in the data model, control flow, and error handling when the model is core.