Lab
Engineering on our own account.
Alongside client work we run research projects that stretch the same skills: machine learning on messy real-world data, consented financial data, and systems where getting it wrong has consequences. They keep us sharp and they occasionally become products.
-
D-02 / V-03 · Financial health from consented data: working footage or sketch 01 · Active, in production with clients
Financial health from consented data
Our longest-running line of research: using machine learning and, increasingly, language models to understand a person’s or a business’s financial life from Open Banking data, so that lending and advice decisions can be fairer and faster. Much of what we learn feeds directly into client work.
-
D-02 / V-03 · Open Banking for accountancy workflows: working footage or sketch 02 · Prototype
Open Banking for accountancy workflows
Tooling that connects consented transaction data to bookkeeping and reconciliation, applying our categorisation engine to a problem every small business has. Exploring whether an accountant’s month-end can be mostly automated without losing judgement.
-
03 · Research
Machine learning for rider safety
Applying our sensor-data and on-device machine-learning skills to motorcycle safety: understanding what is happening around a rider and warning them in time. Deliberately outside finance, because the engineering problems (noisy data, real-time inference, no room for false confidence) are the same ones we care about everywhere. May become a venture in its own right.
Why it matters
From Open Banking to intelligent, connected finance.
Open Banking started as a competition remedy and became the template for consented data sharing worldwide: Open Finance, Smart Data, digital identity. The next step is intelligence. We apply machine learning and AI to that data to understand people and businesses properly, and we think that is the most useful thing a financial-technology engineer can do right now.
-
Open Finance and Smart Data
The Data (Use and Access) Act 2025 created the framework for smart data schemes beyond banking, and the FCA has set out open finance as a priority. The patterns are familiar to anyone who implemented Open Banking: directories, consent, security profiles, conformance. We have done that work once already.
-
Machine learning on Open Banking data
Consented transaction data describes a household or a business far more accurately than a bureau score, but only once a machine can read it. Categorisation, income and expenditure modelling and affordability analysis are where we have spent the last five years, and where we see the most value for lenders, advisers and the people they serve.
Working on a hard data problem? We like those.
If you have consented financial data and a decision to make from it, or a research problem that needs production-grade engineering, talk to us.
hello [at] webovate [dot] co [dot] uk How to get in touch