Setive: preparing Windows and WSL without overwriting a developer's environment
Pure planning, idempotency, real-state verification, and safe undo: Setive's engineering decisions and documented testing limits.
Blog
Applied software engineering
Technical notes on architecture, digital products, AI, and software delivery.
Subscribe via RSS
Fiz Meu Site: when a static product is better than building another SaaS
How a Next.js configurator turns answers into an exportable static website without a mandatory backend, while preview and production share one source of truth.
AI Shouldn't Handle the Money: Modeling RetomaVenda with Ledgers, Leases, and Idempotency
How to surround a probabilistic component with deterministic invariants for credits, privacy, idempotency, leases, and cost in an AI product.
ScriptC: native TypeScript without Electron, Node, or V8 — where it actually fits
ScriptC compiles TypeScript to native binaries through LLVM and only embeds a JavaScript engine when you opt into dynamic mode. I looked at where it fits, what it does not replace, and where I would try it.
Does Clean Architecture still matter when AI writes the code?
Coding agents made code cheaper to produce. That did not make architecture, readability, and testing less important — it changed where those principles need to be enforced.
Kimi K3, one million tokens, and the context trap: more memory is not more architecture
Kimi K3 and K2.8 push coding agents toward huge context windows. Useful, yes—but loading an entire repository is not a substitute for retrieval, tools, tests, and a good harness.
In 2026, the model is only one component: the agent architecture I would actually run in production
Kimi, Jev, the Agents API, and modern coding agents make one thing clear: differentiation is moving from the standalone model to the harness, tools, controls, and verification.
Jev, decisions instead of strings: why not every agent step needs an LLM
Jev proposes a different layer for agents: typed decisions, probabilities, and confidence. The point is not replacing LLMs, but stopping the habit of using them everywhere.
MCP in production in 2026: I stopped thinking about tools as functions and started thinking about boundaries
What changed in how I understand MCP: stateless protocol design, authorization, tools, private environments, Tasks and why connecting an agent to real systems is an architecture and security problem.
Enterprise AI without sending data outside: the architecture I am using to turn Excel into evidence
How I designed a local pipeline for sensitive corporate data: irregular Excel files, normalization, local AI, verifiable rules, human correction and Power BI without making a chatbot the source of truth.
Why I am learning Python for AI without abandoning TypeScript
After years of JavaScript and Node.js, I added Python and FastAPI to my AI stack—not because of hype, but because different ecosystems deserve different boundaries.
From backlog card to pull request: what I learned building a real software agent
The architecture and trade-offs behind DevAgent: an agent that reads a task, changes code, runs checks, and delivers a pull request humans can review.
From dream to graph: what I learned modeling goals, progress and collaboration with React Flow and Three.js
How I built Praxis: one goal tree rendered as a 2D map, list and 3D constellation, with derived progress, cycle prevention, collaboration and optional AI.
SEO without promising a cure: how I built a psychology website that needs to convert within ethical limits
What I learned building Flora Psicologia: technical SEO, content without sensationalism, simple scheduling, sensitive data and an architecture designed around trust instead of aggressive conversion tactics.
AI in compliance without outsourcing the decision: what I learned building a PIF SaaS for European cosmetics
How I modeled Eurologado: AI for drafting and organizing, structured rules for validation, versioning, evidence and audit trails in a regulatory domain.
When the floor plan becomes the interface: how I built a private architecture portal around context
Behind the Estúdio Dupla Face portal: interactive floor plans, responsive coordinates, RLS, private documents with signed URLs and a public site that can survive database failures.
How do you sell something that does not exist yet? 3D preview, approval and architecture in ImpressãoMais3D
How I modeled an on-demand 3D printing store with browser preview, auditable approval, modular boundaries and tests focused on the irreversible risk of physical production.
Variable pricing, artwork and production: how I modeled a print-shop e-commerce without turning everything into exceptions
How I built Digital Tech MS: server-side configurable pricing, quotes, immutable order snapshots, artwork approval and a production flow that treats physical mistakes as business risk.
Monitoring e-commerce is not just pinging: how I built PulseWatch and a reliable alert pipeline
Behind PulseWatch: health checks, state transitions, thresholds, multi-channel alerts, multi-tenant isolation and the decisions that keep monitoring from becoming noise.
From monolith to Kafka + ClickHouse: how I redesigned analytics for millions of events a day
What I learned migrating a marketing analytics SaaS to an event-driven architecture with Kafka, Go, ClickHouse and PostgreSQL without stopping the product.
A portfolio is not a résumé: how I turned lagrotti.dev into an engineering product
How I designed my portfolio to prove engineering decisions, tell real project stories, and build trust before the first conversation.