01
AI isn't always the answer.
We'll tell you if a database index, a rule-based system, or a simple heuristic is a better fit. We have no incentive to overcomplicate things.
// DANIELE & GIULIA
AI where it works.
Engineering where it doesn't.
Two senior engineers. We left enterprise consulting to work on problems we find interesting, with people we like. AI is part of our toolkit — not the whole story.
// HOW WE THINK
A few honest things.
02
We charge for outcomes, not hours.
We left consulting because it punishes efficiency. If we get better and finish faster, that's the point — not a reason to pad the timeline.
03
We work with people, not org charts.
We take on two or three projects at a time. We want technically honest clients — people who understand that AI doesn't produce deterministic results.
// WHAT WE DO
Two ways to work together.
We build full systems. AI where it makes sense, something else where it doesn't.
Find out if your idea is worth building.
A scoped engagement — usually one to two weeks — to assess technical feasibility, realistic cost, and whether AI is actually the right tool. We'll give you an honest verdict. If the answer is don't build it, that's the deliverable.
Deliverables
Feasibility report · Tech stack recommendation · Cost estimate · Build / don't-build verdict
We build the whole thing.
Backend, frontend, AI pipelines, infrastructure, observability. No handoffs, no junior developers, no outsourcing. We work from architecture to production deployment. You own the code when we're done.
Deliverables
System architecture · Full-stack implementation · Production deployment · Monitoring & observability
// SELECTED WORK
Things we've built.
E-COMMERCE / AI
LLM Agent Reliability
An AI e-commerce startup had agents that worked in demos and broke in production. We rebuilt their observability layer, traced failure patterns, and got reliability and latency to an acceptable level.
LLM agents · Observability · Performance engineering
FINTECH / AUTOMATION
Tax Return Automation
A US VC-backed startup needed to automate individual and business tax returns across professional accounting software. We built the full pipeline — OCR, computer vision, reliability infrastructure, and a screenshot-based debugging layer. Six months to production.
OCR · Computer vision · Process automation · Observability
NONPROFIT / SEARCH
Semantic Book Search
A benefit association needed to search across a large corpus of books to find words and passages. We built a semantic search system using Solr. No LLMs — classical search was the right tool for this one.
Semantic search · Solr · NLP
// WHO WE ARE
Daniele & Giulia
Two senior engineers. We spent years in enterprise consulting — long enough to understand how large organisations work, and why we didn't want to stay.
We left because we wanted to manage our own time. And because we couldn't accept a model where getting faster at your job just means more work, not more freedom. If we get better and finish in two hours, we're done for the day. That's the point.
We take on two or three projects at a time. We're not building an agency. We want interesting technical problems and people we'd enjoy having a coffee with.
// NOT FOR EVERYONE
We work well only with people who are willing to collaborate for real — clear requirements, good data, decisions made together. If you're looking for someone to hand a problem off to, we're not the right fit.
Daniele
Senior Software Engineer
AI systems and backend engineering. Builds the pipelines, evaluates the models, and makes sure the whole thing holds in production.
LinkedIn →
Giulia
Senior Software Engineer
Fullstack engineer. The kind who can debug a query, fix the API, and ship the UI in the same afternoon.
Blog →// NOT SURE YOU SHOULD BUILD IT?
Find out before you commit.
We offer a standalone technical assessment — one to two weeks — to tell you honestly if your idea is worth building, what it will cost, and whether AI is the right tool.
WEEK 0
Scoping
We understand the problem, the data, and the users. We might have questions that already change the scope.
WEEK 1
Assessment
Technical feasibility, stack recommendation, cost model, honest verdict.
WEEKS 1–4
Build (optional)
If the assessment says go, we build it. Same team, no handoff.
€3k – €5k
Validation Only
Feasibility report · stack recommendation · cost estimate · build / don't-build verdict
€15k – €25k
Validation + Build
Full assessment + working MVP deployed to production · source code · monitoring
// WRITING
How we think about engineering and AI.
// GET IN TOUCH
Tell us what you're building.
One email. No discovery calls, no decks. If it's a fit, we'll know quickly.
Write to us
hello@41dots.com
We respond to every message. We'll tell you honestly if we're not the right fit.