From scattered data to clear decisions
The information already exists. It is spread across accounting software, spreadsheets and email, and answering a simple question takes days. We build the systems that put it within reach.
What we build
Dashboards and reporting
One place to see margin, cash flow or sales without rebuilding the spreadsheet each time, kept up to date on its own.
Data integration
Connecting accounting software, point of sale and spreadsheets so they stop telling different stories. Python, dbt and SQL on cloud databases.
Automating manual work
The report someone assembles every Monday, the reconciliation that takes half a day: processes that run themselves and flag it when something does not add up.
AI where it can be checked
Pulling data out of invoices and documents, classifying transactions, summarising. Only where the answer can be verified against the source.
How we work together
A fixed-scope build: a working system, delivered and explained. Not open-ended advice, and not a report full of recommendations — something that is left running.
-
First conversation
No cost
You tell us what you can't see in your data today, and we tell you honestly whether it is something we should be solving.
-
Week 1 — assessment
Charged separately, whether or not you continue
We look at where your data actually lives today and what is being done by hand. It ends one of two ways: with a confirmed scope and price, or with the assessment handed to you in writing and nothing further.
-
Week 2 onwards — the build
Fixed price, agreed before we start
What we agreed gets built, to a timeline set at the end of week 1 against the real scope — not a number invented before looking at your data. It is delivered working and explained, so you can maintain it without depending on us.
You can stop at the end of week 1, no explanation needed. The assessment is yours regardless.
Why us
-
Five years building data systems Juan Felipe Salcedo, data engineer with an MSc in Statistics and Data Science. He has worked across very different industries — customer loyalty and marketing at Leal, in Colombia, and today industrial manufacturing at a global company — and that range teaches something concrete: every business measures different things. The data model comes from the business, not from a template.
-
We translate needs, we don't receive specifications Almost nobody arrives with a written requirement. They arrive saying “I don't know whether this product is actually making me money”. Turning that into a data model is the hard part, and it is what we do every day.
-
Rigour before novelty We build AI systems that cite their sources and score how well grounded each answer is, so it is obvious when the model is speculating.
AI you can check
The most common objection to using AI with financial information is that it makes things up. It is a fair objection and worth taking seriously.
So the systems we build return the source alongside the answer, plus a measure of how well supported it is. When the model has nothing to stand on, that shows. For numbers that end up in a tax filing or a business decision, that difference is not a technical detail.
What we don't do
- We don't develop custom applications or websites.
- We don't replace the accounting work. This builds on it.
- We don't propose AI when a well-built spreadsheet solves the problem.
- We don't take on work that needs immediate support around the clock. We work to a defined scope and agreed timelines.