Frequently asked questions
The questions we are actually asked, answered without hedging. If yours is not here, write to us: the team answers directly.
What formats does it take?
Price lists as they actually arrive: spreadsheets, CSVs, PDFs, email attachments. The point is not the format but the shape: no two suppliers use the same headers, and almost none has a "product type" column. That is why KOSMOS works from the descriptive text of the rows instead of demanding a fixed layout.
How accurate is the AI?
We do not publish a percentage, because it would be a percentage on which files. What we can tell you is how the system is built so that accuracy matters less: AI only handles what the dictionary and the learned rules have not already settled, every proposal it makes is visible and editable, and a human correction becomes a rule that holds for every later import from that supplier.
What happens when it gets something wrong?
You correct it, and the correction stays. Proposals are not applied silently: they are visible, they carry the mark of what made them, and a manual edit always takes precedence — even over a whole cluster that was already resolved. You do not fix the same mistake twice, because it becomes a rule for that supplier.
Where does the data I upload go?
Into your own client’s space inside the EurekaGO infrastructure, where it is used to produce your normalized records. The rights over what you contribute stay with whoever generated it: that is written into how the system works, not offered as a commercial promise. The processing detail is in the privacy policy.
Does it work with our catalogue rather than yours?
Today KOSMOS validates against the data model and the ontology of the EurekaGO suite’s cycling vertical. The logic — a resolution cascade, rules learned per supplier, a human who can overrule, no automatic writes to the master — is not specific to cycling, but every sector has its own idea of what a product is, and that ontology has to be built with people who know the trade before there is anything to validate against. If that is your case, let us talk.
Do we have to adopt the whole suite?
KOSMOS is a pipeline platform: it needs a product model behind it, because resolving against that model is exactly what it does. So it does not run isolated from the catalogue. The rest of the suite — marketplace, commerce, warehouse — is independent of that choice.
Where do we start?
With a real file. Send us a price list as you receive it, with all of its oddities, and we will tell you honestly what would come out of it and where a human decision would still be needed. That is worth more than any demo on clean data.
The most useful file is the messiest one.
Send us a price list exactly as it reaches you. We will tell you what would come out of it, and where a person would still be needed.
Let’s talk →