About Us

Anthony Cosgrove MBE and Dr Sebastian Drave met at HSBC during a period of intense scrutiny.

The bank had just been fined USD 1.9bn for serious AML failures, and large parts of its financial crime capability had to be rebuilt.

Anthony was responsible for delivering the core technology inside the Financial Intelligence Unit, a specialist team focused on the most complex and high-risk financial crime. At the same time, Seb became HSBC’s Chief Data Scientist for Financial Crime, building the data science function and applying machine-learning techniques to uncover criminal behaviour that rules alone could not detect.

In 2016, they helped deliver the Financial Intelligence Platform, a large-scale data environment containing billions of transactions. It enabled new analytical approaches, but also exposed a persistent problem. Testing relied on production data by default, even though it is legally sensitive, biased by historic decisions, and has limited coverage of confirmed outcomes. Teams could change systems, add models, or switch vendors, but still lacked objective evidence of whether detection was actually improving. The absence of a reliable feedback loop became one of the most frustrating constraints they encountered.

After leaving HSBC, Anthony co-founded Harbr Data, a platform designed to enable secure data collaboration. Seb later joined him, and together they worked closely with the UK’s Financial Conduct Authority. For the FCA’s Financial Crime TechSprint in 2019, they provided synthetic transaction data with embedded criminal behaviour, allowing teams to build and test AML solutions without using real customer data. Following the event, technology companies began using the data to test and train their systems, and it became clear that synthetic data could become the missing feedback loop.

Seb rebuilt the early prototype using agent-based modelling techniques, proprietary input statistics, and his deep financial crime expertise. The result was a national-scale synthetic financial ecosystem with hidden criminal networks and realistic money laundering behaviours. For the first time, teams could test AML systems end-to-end and objectively measure detection performance. What started as a dataset evolved into an AML Testing Platform used for baseline assessments, system optimisation, vendor comparison, and model development.

Today, that need is even more acute. AML systems are increasingly powered by AI and machine-learning models that are adaptive, probabilistic, and highly sensitive to data distributions and feedback loops. Apparent improvements can be misleading, and production data provides little clarity on what systems detect, what they miss, or how performance shifts over time. Syntheticr exists to solve this problem. By combining high-fidelity synthetic data with objective performance scorecards, Syntheticr restores the missing feedback loop and gives teams clear evidence of how their AML systems actually perform. All without relying on production data.