Openlayer, which claims to be the creator of the world’s most comprehensive platform for testing AI, announced that it has closed a $4.8 million seed round.
The funds from the investment round will be used “to expand its workforce and enhance the platform’s functionality so it is capable of handling additional machine learning (ML) tasks.”
Already, startups to Fortune 500 companies are reportedly using Openlayer “to test and validate their machine learning models, uncover unexpected mistakes, and diagnose why and when they’re happening.”
Rishab Ramanathan, co-founder of Openlayer, said:
“As ML becomes more popular and more accessible, it’s critical we put the right guardrails in place for models entering the real world. Testing should be a vital part of building ML models from the start, as opposed to as an afterthought. Such test-driven development is the only way to meaningfully align models with human interests. Doing this in a repeatable way requires a platform that makes battle-tested models a reality.”
Openlayer’s founding team is “composed of former Apple ML engineers who have firsthand experience building AI at scale.”
Other members of the company “include an ex-Amazon engineer and a Harvard Design Engineering school graduate.”
Vikas Nair, co-founder of Openlayer, said:
“Over the course of our time working across 15 different teams at Apple, we faced firsthand the same problem over and over: there is no standardized way to test and collaborate on machine learning models. Because of this, errors are often not caught until after shipping, which can result in potentially disastrous outcomes.”
While numerous software testing platforms exist, those platforms are designed “for deterministic systems in which a given input will produce an expected output.”
Since ML models “are probabilistic, there has been no way to test them reliably until now.”
Openlayer helps teams systematically improve their models and datasets by:
- Verifying the integrity of training and validation datasets
- Surfacing meaningful discrepancies between training, evaluation and production data
- Ensuring models meet their target performance benchmarks
- Validating models are robust to edge-cases by generating synthetic data to inject noise and conduct adversarial attacks
- Guaranteeing fairness of model behavior across data subpopulations
- Tracking versions of models and datasets and comparing their performance
- Explaining model behavior by surfacing which features of the data were used to make a prediction
The seed round will “enable the Openlayer team to create more sophisticated guardrails for customers to test their models against as they iterate.”
The platform will also “allow for edge-case detection using synthetic data to generate test cases they might not have considered.” Importantly, customers will “benefit from faster, more organized development velocity.”
Gabriel Bayomi Tinoco Kalejaiye, co-founder of Openlayer, said:
“Over the long term, we envision a future in which many of the processes for detecting and fixing errors can be automated with Openlayer. Our vision is to be the go-to hub for any ML team shipping models. To achieve this requires building a development pipeline that delivers powerful insights about your models and data every step of the way, pre- and post-deployment.”
Astasia Myers, enterprise partner at Quiet Capital, the lead investor, said:
“AI adoption is on the rise, and we are seeing the importance of data-centric ML as algorithms become commoditized. The Openlayer founders worked on ML at Apple and saw firsthand the benefit of having data-centric ML solutions that supported test-driven development and data quality analysis. The founders took their unique insight and applied it to building Openlayer that solves this need for all businesses. They are tackling a critical problem around ML data intelligence which we expect will continue to boom with AI’s increased ubiquity.”