1. General

Category

Category

Savings & Loans

2. Project Details

Company or Institution

DreamQuark

Project

Sustainable investment Brain

General description of the AI solution

DreamQuark offering consist on an advisor portal for investment advisors to give them a 360° view on customer's assets as well as their preferences. This view is leveraged by explainable AI algorithms to give them access to next best action recommendations. For each security we provide ESG KPIs to make recommendations to improve these KPIs. This portal leverages all the expertise put into Brain our trustworthy AI operationalization platform that enables citizen data scientists (and whoever without AI and ML expertise and coding) the capacity to build, deploy and maintain trustworthy AI models (with model documentation, fairness, explainability, robustness assessment and ROi measurement functionalities). We then provide a last tools that enables the collection of data and the orchestration of data (databases, files or APIs) and ML models (either built with Brain or available in open source) to feed our portal with models that represent one customer context. All models and functionalities are available through SDKs as well. This solution is highlighted by several firms including Gartner as a way to reduce ethical risks posed by AI

Website

Https://www.dreamquark.com

Organisation

DreamQuark

3. Aspects

Excellence and Scientific Quality: Please detail the improvements made by the nominee or the nominees’ team or yourself if your applying for the award, and why they have been a success.

DreamQuark leverages explainable AI, including for deep-learning and transformers, auto-ml and mlops to enables organization to build, deploy and manage advanced AI applications. Gartner has made several mentions of our solution as a responsible AI cool vendor and the Monetary Authority of Singapour with Accenture have stated "After assessing proposals from FinTechs, AI providers, and financial institutions from across the world, our team thought that your solution was unique and showed great potential in implementing Responsible AI to guard against bias and provide fair access to financial services" as we have been finalist on the Global Veritas Challenge.
DreamQuark IP (including applications filled for patents) includes : technology for identifying and monitoring bias, explainable AI (NLP, deep-learning and standard ML), NLU and NLG applications. We are developing explainable AI with a focus on end-users and end clients including the development of domain specific explainations (leveraging knowledge graphs and domain ontologies) as well as plain text automatically generated explainations. On top we are developing functionnalities to enable the integration with Gaia-x architecture (as part of a trustworthy AI initiative) as well as technics to control the ecological impact of ML applications and our CEO is part on a taskforce around the European Social Taxonomie.
Applications include : financial inclusion as well as the reduction of CO2 emissions of securities portfolios by the financial services leveraging AI

Scaling of impact to SDGs: Please detail how many citizens/communities and/or researchers/businesses this has had or can have a positive impact on, including particular groups where applicable and to what extent.

DreamQuark specifically integrates ESG data as part of its solution and we both provide reporting on the current emissions of ones investment portfolios but also leveraging explainable AI (product recommendation) to recommend securities that would reduce the overall emission of the portfolio. We can therefore clearly measure the overall reduction of CO2 occured by our solution. DreamQuark also provides bias reduction technologies and we are working with banks to help them overcome their challenges while deploying these more ethical and fair AI solutions. We could measure the number of persons that have gotten an access the credit that would have forbidden without this type of technology.
We also recommend the reading of our whitepaper (state of Ethical AI) where we specifically mention the SDG. We currently sell our solution in Europe, Singapore (and by extent south east Asia), India and North America. We have a distribution partnership with Atos on this solution.

Scaling of AI solution: Please detail what proof of concept or implementations can you show now in terms of its efficacy and how the solution can be scaled to provide a global impact ad how realistic that scaling is.

We are currently deployed in a couple of Banks and are engaging into new Proof of value prior to a larger deployment. We have signed a distribution partnership with Atos and are developing a partnership with Netapp and other software providers. We are finalists of the Global Veritas Challenge. The solution is production ready, can be deployed accross AWS, Azure, IBM, GCP and we are engaging with other cloud providers. We provide privacy and production oriented features to help scale.
We are working with Gaia-x to provide compatibility with Gaia-x architecture.
We have raised 20M to date to develop the product and deploy the solution. To date we have scaled communication and automated our marketing approach. We envision to give access to this solution directly on our website to ease the scalability of our model. We have published open source implementations of explainable deep-learning models that have received praised (tabnet)

Ethical aspect: Please detail the way the solution addresses any of the main ethical aspects, including trustworthiness, bias, gender issues, etc.

DreamQuark solution provides trusworthy AI on two major dimensions. First we provide through an advisor portal a way to integrate the ESG preferences of ones client in order to reduce the overall CO2 emissions of securities portfolio through the redirection of investment towards more responsible companies. South countries are the most affected by the climate change and such a solution could help stay on a 1.5° to 2° trajectory. Second we provide robustness functionalities (to spot errors, reduce data leakages and monitor data and concept drifts or PSI) as well as equal opportunitied (the user selects the variables to be protected and our algorithms learn during the process to rebalance the algorithm to have the same score distribution for all values of these protected variables with a minimal drop in model performance. We do that for either demographic parity or equalized odds metrics).
Our offerings is in compliance with disclosure, fit for 55, MIFID II, the french article 29 regulations on ESG as well as GDPR (we offer tools to obfuscate sensitive data, data security functionalities as well as secured data processing processes) as well as the new regulation on trustworthy AI. Our offering provides all aspects of ethics (we also have a PhD student in ethics working on causality, explainability and fairness). Finally our tagline is we make AI simpler, smarter, fairer. All our development principles are in our whitepaper on the state of ethical AI