Request A Demo

FREQUENTLY ASKED QUESTIONS

FAQs

While the SolasAI bias mitigation functionality is a form of GAI, it is not a large language model, it is not trained on public data, or share data outside of each mitigation process. It is entirely self-contained and designed in an air-gapped environment.  The SolasAI algorithm creates alternative models only using the customer’s model, model results, model metadata, and model development data. All of this information is entirely self-contained in a given analysis. The algorithm does not learn from any model or dataset other than the one that it is using for a given project.

No. Each run of SolasAI only learns from the specific model and data that the SolasAI software is given access to for a given run. 

SolasAI uses the customer’s inputs to test, explain, and generate alternatives without reusing or sharing this information. 

Each alternatives run is a unique operation.

  • Life, P&C, and health insurance underwriting 
  • Fraud detection
  • Risk scoring 
  • Pricing 
  • Marketing 
  • Automated valuation models 
  • Credit underwriting 
  • Clinical decision support models 
  • Small business lending
  • Employment - resume screening and interview bot

No, SolasAI rarely requires changing the model architecture or type of algorithm. We rarely find this provides any advantage or is worth the extra effort. 

SolasAI is designed to test, explain, and optimize the model you provide.

SolasAI is a set of Python libraries architected to run in standard Python environments either on-premise, in a customer’s cloud environment, or as a shared service. 

SolasAI provides access to the software in multiple forms including Docker image, cloud machine image, downloadable Python libraries, or Python wheels (preconfigured installation files.) We provide minimum specifications and recommended architectural elements.

It tests for disparities across protected groups using conventional and well-established metrics within the banking, insurance, fintech, and healthcare industries like the adverse impact ratio and the standardized mean difference.

The SolasAI software uses iterative multi-objective learning algorithms to construct alternative models that optimize performance while reducing disparities.

Absolutely. SolasAI uses sophisticated explainable AI to pinpoint features or decisions that may be contributing most to disparities—helping you take targeted action.

Yes. SolasAI is designed and tested to stay up-to-date with the latest regulatory guidance, research, and fairness standards.

While SolasAI supports your compliance journey, it complements—not replaces—your broader legal and regulatory responsibilities. SolasAI makes it easy to quantify and understand your fair lending risk so you can focus on how it impacts your business.

We do offer the Disparity & Bias Testing Library for free. It provides everything you need to perform testing. You will need the full paid version for the explainable AI and less discriminatory alternatives modules to uncover and mitigate the underlying causes of any disparate impact or bias.

The free library can be found here: https://github.com/SolasAI/solas-ai-disparity

SolasAI relies on the customer’s access management, authentication, or container management system. 

There is no concept of a “user” in SolasAI, nor of logging into the platform as these are Python libraries or Docker containers that run in the customer’s infrastructure. 

The customer is expected to secure whatever environment they are installing SolasAI into using their existing access management, authentication, or container management systems. 

SolasAI only requires credentials for downloading the libraries, Docker images, or updating the shared installation of SolasAI.

SolasAI can run completely isolated from the internet.

We do have ways of helping customers install it without an active internet connection. 

We do find it easiest to install it with an internet connection active for the installation, but it is not a requirement. The connection can then be disabled for production usage. 

SolasAI and its dependencies can be updated through the customer’s existing automation for scanning, updating, and managing Python libraries and other applications.

SolasAI has completed and is able to provide a SOC 2 Type 2 audit report. 

We also have a standard SIG Questionnaire available upon request

No, unlike most of the competitive solutions, SolasAI does not use it directly. Our optimization finds high-quality alternatives that are also fair without having to directly factor protected class status into the alternative models that are built.

If information for protected class status is not collected, it is standard practice to use the Bayesian Improved Surname Geocoding (BISG) methodology to estimate a person’s protected class status. While getting the BISG values does require knowing the person’s home address and name, SolasAI does not use that information. Instead, it operates only on the BISG values provided to it.

SolasAI partners with experts to provide services such as: 

Bespoke Report Fee & Expert Consultant Review: These are custom reports that build on the analysis generated by SolasAI and written for a specific external stakeholder or type of external stakeholder. This could

include writing reports to the specific requirements of a particular partner bank, group of banks, auditor, or specific regulatory authority. It also receives expert validation and written commentary from an expert consultant from one of our expert partners. This is a deep review of the results and does provide support for communicating the results achieved. 

Experts to Perform Testing & Alternatives Analysis: Some SolasAI customers request to have experts perform the analysis. This may be due to high demand, a particular model being very high risk, or a shortage of skills. For SolasAI subscription members, an expert partner consultant can perform the testing and LDA search for the customer. This is a one-time charge or can be built into the subscription price itself. 

Data or Model Governance and Management: SolasAI partners include deep expertise in developing the policies and procedures for effective data and model/AI governance. They provide a wide range of services directly in support of the SolasAI software or more generally for data quality, governance, and usage.

Most teams can begin seeing insights quickly—often within days—since the software is designed to integrate into your environment and existing systems.

SolasAI is designed for ease of use, but we also provide onboarding support and training for your team to get the most out of it.

Clients typically see measurable reductions in disparity, enhanced model transparency, and increased confidence in their decision systems.

SolasAI is often used by modeling teams, dedicated compliance analysts, and consultants to perform the disparity testing and mitigation. Then reports and other output are generated to help these stakeholders and decision-makers choose the best options for their business.

SolasAI is offered as either a Team or Enterprise subscription. First, a Base Subscription is purchased that covers the compliance team, all the startup activities, and ongoing support. Then, the customer adds additional subscriptions if they decide to roll it out to a wider group of modelers. Please contact sales@solas.ai for more information.