Taxonomy alignment first
Before any model runs, we map our categories to your chart of accounts. Output lands directly in your reporting structure without a second translation step.
zuqytyi - Dublin, Ireland
Expense data
Expense data
that makes sense
We built Zuqytyi because financial teams were spending hours sorting transactions that software should handle automatically. Our focus is narrow on purpose: AI-driven expense categorisation, done with enough care that the results are actually trustworthy.
What the work actually involves
Most businesses accumulate transaction data that sits in spreadsheets, half-labelled and inconsistently tagged. Accountants reconcile it manually, finance managers second-guess the categories, and audits slow to a crawl. Zuqytyi was founded in 2019 to address exactly this gap - not with a generic automation layer, but with models trained specifically on how businesses actually spend money.
Our system reads transaction descriptions, merchant names, and contextual signals to assign categories with a level of consistency that manual tagging rarely achieves. When the model is uncertain, it flags the item rather than guessing - because a wrong category that looks confident causes more damage than one that's honestly marked for review.
We work with clients across professional services, logistics, retail, and technology. Each engagement starts with a mapping phase where we align our taxonomy to the client's chart of accounts, so the output integrates directly into existing reporting workflows rather than requiring a second round of translation.
Uncertainty is surfaced, not hidden
Low-confidence transactions are flagged for human review. We treat an honest question mark as more useful than a confident wrong answer.
Models trained on real spend patterns
Our categorisation logic draws from actual business transaction data across sectors, not generic financial datasets assembled for benchmarking purposes.
Remote delivery, no location constraint
Our team operates fully remotely with clients worldwide. Onboarding, configuration, and ongoing support happen without requiring physical presence.
The people behind the system
Zuqytyi is a small team by design. We kept it that way because every client engagement involves direct access to the people who built the models, not a support tier. The specialists below handle the core of what we do.
Orsolya Fekete
Head of Model Development
Orsolya leads the categorisation model work, including training data curation and the flagging logic that determines when the system should defer to a human reviewer. She came from a background in computational linguistics before moving into financial NLP.
Tadhg Ó Murchú
Client Integration Lead
Tadhg manages the taxonomy mapping phase at the start of each engagement and handles the technical integration into accounting platforms. He previously worked in finance systems consulting, which shapes how he thinks about what "done" actually means for a client.
Ragnhild Svensson
Data Engineering
Ragnhild maintains the pipeline infrastructure that processes client transaction feeds and routes flagged items to the review queue. Her work is mostly invisible when it goes well, which is the point.
Bartosz Wiśniewski
Finance Domain Specialist
Bartosz brings a decade of management accounting experience to the team. He reviews edge cases where the model's categorisation logic intersects with accounting standards, and contributes to training data quality for sector-specific spend patterns.
