Etiq AI is a flexible software solution for building trustworthy Machine Learning (ML) algorithms that enables transparent and auditable decision-making.
AI pipelines contain anomalies and errors that can lead to algorithms suggesting decisions that negatively affect businesses and their customers. It prevents organisations from addressing AI performance and inclusivity metrics. Etiq AI customers can access our AI out-the-box tests so that they detect and rectify issues of accuracy, drift, bias, fairness at any point of their AI development/deployment pipelines. The Etiq AI platform simplifies the process so that testing happens throughout the pipeline, prevents accuracy loss in production, while reducing the time it takes to validate and transition a model from prototype to production.
Our solution is a suite of ‘plug-and-play tests, available via API access. These simplify the identification and viewing of errors, efficiency and bias metrics on an interactive and unified dashboard.
Etiq AI ensures that different user personas across the business have increased explainability and control over the AI models that they are deploying. We provide users the ability to access outputs via one dashboard for, e.g., to increase communication about algorithmic decisions to customers, explainability about decisions to data scientists, compliance and commercial teams. This makes it easier to report to non tech stakeholders and to regulatory bodies with minimum effort. We seamlessly enable observability for any model in any environment.
By providing data scientists with tools that reduce errors in data and models and placing guardrails on algorithms, the risk of errors, potential flaws and missed opportunities is vastly reduced. Etiq increases data scientists’ confidence in the performance of their models as continuous monitoring is in play so that model production can be deployed faster. We streamline testing, model deployment and monitoring so that data scientists and engineers are able to react faster to mitigating risks. Our solution saves the organisation 10 data scientist man days per test run, while increasing robustness and confidence levels and reducing error rates.
Finally, we ensure that risk compliance teams can account for automated decisions thus meeting regulatory requirements of compliance, fairness and explainability. As explained above, our tests identify the potential for flaws, errors and unintended bias creeping into AI models unnoticed. The tests allow users to report on their AI logic to regulatory bodies in easy to understand terms for non-technical readers. This allows for streamlined audit and compliance.
By ensuring that our clients’ AI pipelines and models are developed and deployed to mitigate against the potential for bias and other harms, errors can be minimised.
Etiq was born out of the Zinc VC mission 2 accelerator program. Etiq firmly believes technology can be used to automate decisions in a scalable and sustainable way, however, in many cases anomalies, errors and unintended biases can impact your algorithm’s accuracy and your business efficiency.
That’s why Etiq has created a solution to give everyone on your team, from business stakeholder to data engineers, visibility to monitor your algorithms are doing what you built them to do and minimise any unintended bias that may have accrued.
Raluca has 10+ years of experience in data science working with a variety of clients (UK retailers, banks & telco companies). Prior to ETIQ she was Director – Data Science for Merkle Aquila.
Raluca is passionate about building solutions to help data teams to build a robust and responsible AI.
Iris initially trained as an accountant/tax advisor in the Hedge Fund industry. Prior to Etiq, Iris built another successful tech start-up, covering functions such as tech, operations, sales and management.
Iris is passionate in equal parts about equality and building successful businesses.