Research Data Collection

Does your research or analytics involve unstructured source data like news, annual reports, and regulatory filings? Is the majority of research time spent on data preparation rather than analysis? Hivemind can help. Hivemind solves this for Winton by splitting such data projects into micro-tasks, which our platform distributes to a specialised team of data annotators. Timescales are dramatically reduced, opening up many new and potentially groundbreaking lines of research.

Read HFMWeek's deep dive on Hivemind at Winton


Winton is a global investment management and data science company. Founded in 1997 by its CEO David Harding, Winton’s business is grounded in the belief that the scientific method can be profitably applied to the field of investing.



Website Data Collection with Distributed Teams

The world of alternative investments such as hedge funds and private equity is well known to be intransparent in comparison to public markets. The little information that is openly available can be scattered across thousands of websites in proprietary formats. Hivemind allows the complex process of collecting such data to be broken down into a sequence of discrete tasks of varying difficulty. By allowing tasks to be routed flexibly between internal and external data collectors located both onshore and offshore, Hivemind can accelerate the overall process while not losing transparency on data quality.


Preqin is a leading provider of data, insights and tools for alternative assets professionals around the globe. More than 60,000 investors, fund managers, placement agents, service providers, advisors and other industry professionals rely on Preqin to find valuable opportunities and make intelligent decisions in alternative assets.



Machine Learning Training Data

With all the excitement about modern AI techniques, it is easy to forget that supervised learning algorithms are only as good as the training data they are fed. But how do you obtain high-quality training data? Online crowdsourcing only works for the most trivial tasks, while traditional outsourcing often takes too long to put in place.

Hivemind enabled Ripjar to turn thousands of raw text extracts into a carefully crafted training dataset for its AI powered threat detection systems. A task pipeline fusing NLP with human annotation maximised throughput, while maintaining quality. Our native support for combining internal, outsourced, and crowdsourced project contributors ensured that the right skills could be quickly sourced and scaled up.


Ripjar is a global company of talented technologists, data scientists and analysts designing products that will change the way criminal activities are detected and prevented.



Crowdsourced Image Annotation

Imagine you run an innovative company that's big on cutting-edge technology but short on time and manpower. With 200,000+ images that need careful labelling for a new computer vision product, you calculate it would take a new recruit 2 years to complete the task!

Hivemind's seamless integration with Amazon Mechanical Turk allowed Hoxton Analytics to instantly crowdsource its project to up to 1,000 distributed workers, who completed the 2 years of work in just a few days. Our integrated data-quality controls ensured data integrity, by automatically focusing crowdworkers' efforts on the trickiest tasks.


Hoxton Analytics' market leading technology provides uniquely rich retail intelligence. Through floor-level sensors using computer vision, their technology counts visitors and derives behavioural characteristics without gathering personal information.