The FDE Stack: Tools for Embedded Engineers
One backend language, Postgres, one LLM API, and a deploy target the client can pay for. The FDE stack is boring on purpose, and boring wins onsite.
Read article → Tools & StackLocal-First Apps for Field Work
Warehouses and farms have terrible connectivity — here's the local-first sync architecture that keeps field software honest even without the cloud.
Read article → Tools & StackIntegrating With Client ERPs: A Survival Guide
ERP integration is political archaeology, not just coding. Here's how forward deployed engineers survive SAP, NetSuite, and the systems vendors forgot.
Read article → Tools & StackSyncing Spreadsheets With Systems
Spreadsheets refuse to die. Here's the sync architecture that lets them coexist with real systems without the Friday-night copy-paste ritual.
Read article → Tools & StackCron Jobs and Queues: The Boring Backbone
Background jobs and queues small business: the progression from crontab to production queues, and why boring infrastructure wins.
Read article → Tools & StackTailwind for Speed: Why FDEs Standardize on It
One engineer, a blank CSS file, and a demo tomorrow — why FDEs choose Tailwind when speed matters more than taste.
Read article → Tools & StackPlaywright for Business Automation: No API? No Problem
Playwright browser automation lets you script workflows even when vendors refuse to build an API. Here's how FDEs use it safely.
Read article → Tools & StackLightweight ETL: Moving Data Without a Data Team
dlt, Python, and Postgres — the lightweight ETL stack one engineer can run without hiring data engineers.
Read article → Tools & StackMonitoring Small Apps on a $0 Budget
The error started Tuesday; the email arrived Friday. Health checks, error alerts, job heartbeats, and one status page: the $0 monitoring stack.
Read article → Tools & StackSelf-Hosted LLMs: When Privacy Demands It
Open-weight models on modest hardware genuinely cover extraction, classification, and summarization. Here's when self-hosting is right, and the honest limits.
Read article → Tools & StackVercel vs. Cloudflare: An FDE's Deploy Decision
Both platforms are good; that's what makes the choice annoying. Framework fit, egress pricing, and the operator after month six decide it.
Read article → Tools & StackNext.js for Internal Tools: When It Earns It
Next.js for internal tools is the right call about a third of the time. Here's the decision matrix that tells you which third you're in.
Read article → Tools & StackAuth for Internal Tools: Don't Roll Your Own
Magic links, Google sign-in, or a network allowlist: three auth setups cover about 95% of small-company internal tools. Skip the password tarpit.
Read article → Tools & StackDocker on Client Hardware: Deploy Onsite
How to deploy Docker on a client's own hardware without regret: a three-container stack, tested backups, tagged updates, and honest limits.
Read article → Tools & StackEscaping Airtable: The Migration Playbook
Airtable got you here, but the ceiling hits at 50k rows and three power users. Here's the phased migration to Postgres that keeps the business running.
Read article → Tools & StackSQLite for Prototypes and Small Apps: Stronger Than You Think
No server, no credentials, no 2 a.m. pages. Why SQLite carries a prototype further into production than you'd guess, and the honest signs it's time to leave.
Read article → Tools & StackStreamlit Dashboards: Ugly, Fast, Effective
A weekend-built Python dashboard beat a six-week React quote. When Streamlit is the right call for internal tools, and when it's malpractice.
Read article → Tools & StackChoosing an LLM API: A Builder's Field Guide
Benchmarks won't pick your LLM. Task shape, cost at volume, latency under load, and data policy will, plus the twenty-example eval you can run this afternoon.
Read article → Tools & Stackn8n vs. Custom Code: Where Workflow Tools End
n8n is a great two-week win and a scary place to keep your core business logic. Where the complexity ceiling hits, and the hybrid pattern past it.
Read article → Tools & StackRetool vs. Custom Build: The Honest Comparison
Retool wins the first month and often loses the second year. Here's the cost curve, the hidden taxes, and the hybrid play that actually works.
Read article → Tools & StackSupabase for MVPs: Backend in an Afternoon
Schema at 10 a.m., auth by 2, client poking a live prototype by 4. Why Supabase is the default MVP backend — and the three cases where it isn't.
Read article → Tools & StackPython Automation: The FDE's Swiss Army Knife
Glue code, data munging, scheduled jobs — the unglamorous Python patterns behind most FDE quick wins, and when to put the laptop down.
Read article → Tools & StackPostgres for Everything: The Default Database
Your architecture diagram has five datastores and twelve users. Why boring old Postgres, JSONB and pgvector included, is still the default answer.
Read article → Tools & StackCloudflare Workers for FDEs: Ship at the Edge
Why embedded engineers keep choosing Cloudflare Workers for client builds: no DevOps surface, seconds-fast deploys, and bills a CFO actually likes.
Read article →