Written by practitioners
Authored by the engineers and leads on the engagement. If a piece cites a number — a latency, a bill, a week count — someone on our team measured it.
Insights
Notes from thebuild floor.
No ghostwritten thought-leadership. These are field notes from real engagements — what broke, what we measured, and the checklists we now run every time.
The library
Read what we learned the hard way
Every piece is written by the team that shipped the work and reviewed by a second senior engineer. First articles publish this quarter.
RAG in production: what actually breaks
Retrieval pipelines that ace the demo fall over on stale indexes, silent chunking drift, and eval suites that never met a real user. We walk through the six failure modes we hit across nine production deployments — and the guardrails that caught them.
A CTO's checklist before a cloud migration
Forty-one questions we ask before we touch a workload — licensing traps, data-gravity math, and the egress line item nobody budgets for. Print it, score your estate, and know your real timeline before a vendor tells you theirs.
The real cost model of a dedicated squad
Rate cards hide more than they reveal. We break down what a 5-person senior squad actually costs per shipped feature — including ramp-up, meetings, and the weeks where nothing visible ships — versus the same scope run through staff augmentation.
Agent evals your board will actually trust
Accuracy on a golden set means little when an agent can call tools, spend money, and email customers. Here is the layered eval harness we run — trajectory checks, cost ceilings, red-team suites — before any agent touches production data.
Strangling a monolith without stopping the roadmap
You do not need a rewrite freeze to modernize. A routing-first strangler pattern let one retail client extract four services in twelve weeks while shipping features every sprint. The exact seams we cut, in order, with rollback plans.
Cutting a cloud bill 38% without touching uptime
Most savings hide in three places: over-provisioned non-prod, unattached storage, and compute bought at on-demand prices for steady-state loads. A line-by-line teardown of a real optimization pass on a fintech AWS estate.
Week 0: how we run discovery so weeks 1–8 do not slip
Discovery is not a workshop with sticky notes — it is where the schedule is won or lost. The five artifacts we insist on before writing code: success metrics, a walking skeleton, access checklist, risk register, and a demo calendar.
Nothing published under this topic yet — first pieces land this quarter. Pick another topic or subscribe below and we will send it when it ships.
Editorial standards
Why these are worth your inbox
Reviewed twice
Every article passes a technical review by a second senior engineer and a plain-English pass. If it cannot survive both, it does not publish.
Usable the same day
Each piece ends with something you can run — a checklist, a scoring rubric, an architecture sketch. Reading time is stated up front and honest.
The Build Notes
One email a month. Zero fluff.
New articles, one tool we adopted, one we dropped, and a checklist from a live engagement. Unsubscribe in one click, any time.
- Monthly, never more
- No sales sequences, ever
- Your address stays with us
Subscribe by email
Opens your mail app with a pre-filled note — we add you by hand while the signup form is being built. First issue ships this quarter.
Let's build
Have something ambitious in mind?
Tell us what you're trying to build we'll come back within one business day with a clear point of view and next steps.
