Scaling AI - in particular, shipping models faster and iterating with the newest data - is top of mind for all of our clients. In order to truly scale an AI system, you need to create a scalable data pipeline, where data flows seamlessly and can quickly be used to build, iterate, and test models. That’s why we’re announcing a new end to end API solution that helps AI teams automate their data pipelines and integrate Centaur Labs data annotation capabilities into AI development life cycles.
We’re introducing new APIs that allow teams to seamlessly import data, create annotation tasks, set gold standards and get labeling results entirely through our API tools. All data and annotation types are supported.
When AI teams begin data annotation projects, they often have data ready to annotate today, and will also have additional data to annotate in the future. Perhaps they’re using data collected as part of a clinical trial, and clinical sites share data with them as it becomes available. Or maybe they are a medical device company, and some of their hospital system customers send them data in monthly batches. Or maybe they’re working with patient generated data, and have a steady stream available through an application. Whatever the circumstances, the team will have access to more data over time.
Before, if you had new data available on a Tuesday, and more on Thursday, you would need to spend time manually importing that data twice in the Centaur Labs platform. The same manual processes would need to be taken to download the final labeled data results as well. Now, with API support, you can programmatically have data sent to the Centaur Labs platform as it becomes available in your Amazon S3 bucket. Once you specify which project and task to add the data to, it will be sent to the labeling network, labeled, and the results available for download via our API anytime.
Teams often want multiple types of annotations throughout the model development lifecycle. Perhaps they’re building a model that leverages multiple data types, or as their model improves and their training initiatives become more targeted, they want additional annotations on the same dataset.
Instead of opening the Centaur Labs platform and manually creating a new task, you can now use our API and create the new task right from where you’re already working. You can specify your task type - whether classification or segmentation - write your task prompt and, if needed, a set of answer choices to be shared with the labeling network, upload Gold Standard examples and assign unlabeled data to the task.
An API is only as good as its documentation. From the start, we’ve built our APIs with robust documentation, allowing for:
Do you want to automate your data pipeline so you can iterate on your models faster? Integrate Centaur Labs data annotation capabilities into your AI development life cycle more easily with our new APIs.
Our new APIs are generally available today to all customers globally.
Schedule a demo to learn more!
Today, we’re getting to know Tom Gellatly, a Centaur Labs co-founder and the VP of engineering!