Unifi has now developed a complicated device for collating environmental, social, and governance (ESG) information from complicated company reviews, by tapping into the facility of the generative AI abilities by a Zindi competitors. After combined outcomes from conventional consultancy corporations, Unifi partnered with Zindi to host a GenAI competition on Zindi, the place options are constructed by a group of expert information scientists. This competitors was particularly designed to handle Unifi’s problem of extracting ESG metrics present in PDF paperwork that embrace each textual content, charts, and pictures.This resulted in revolutionary approaches that considerably improved their productiveness and effectivity.
PROBLEM
Unifi’s major constraint in scaling their companies was the incorrect measures and gradual means of extracting ESG information from PDF reviews. This information extraction is crucial in offering complete insights to their shoppers. Unifi had beforehand regarded to numerous consulting corporations for options however they discovered the outcomes unsatisfactory.
SOLUTION
Motivated to get one of the best outcomes potential and to cut back the time and value of the guide course of, Unifi and Zindi hosted a problem on Zindi’s platform with the target of creating an AI resolution for parsing annual reviews in PDF format to extract predefined ESG metrics. All through the competitors, Zindi guided members by internet hosting a webinar to elucidate the issue in depth and conducting a Q&A session to handle members’ considerations.
The problem was open for 4 months and it was accessible to all expert AI practitioners throughout the globe. After the competitors ended, with the assistance of Zindi’s automated analysis and rating programs, a real-time leaderboard which helps members monitor their progress, the highest 10 candidates have been shortlisted based mostly on their accuracy rating.
Reflecting on the competitors, the director of Unifi said, “Zindi crew struck a great steadiness making ready the competitors in such a means that the submissions have been comparable however the intrinsic challenges of the issue weren’t diluted.”
IMPACTS
With submission of the members’ code, documentation and 10 minutes video explaining their resolution and highlighting iterative learnings, the highest three options have been chosen by JG and the crew at Unifi. Because of these members providing a various vary of approaches, Unifi is now taking every of the options and incorporating the completely different approaches making use of them to the fashions that Unifi has operating in manufacturing to drive the effectivity positive aspects.
“This course of gave us confidence realizing that the options proposed could be judged on not solely fixing the issue in an pragmatic means, but additionally virtually deploy issues,” stated JG Cowper, Director Unifi. “Due to the Zindi crew for serving to us speed up fixing this downside for our shoppers.”
The profitable resolution achieved an accuracy of 53%, marking a significant enchancment over the 48% accuracy achieved by the answer Unifi had beforehand sourced and had been utilizing beforehand.
The partnership between Unifi and Zindi reveals how revolutionary collaborations could be of nice worth. By tapping into Zindi’s community of generative AI consultants, organisations like Unifi can supply a assured resolution for his or her AI pushed issues.
