What's new in Beta2

Beta2 for Gretel.ai is all about delivering privacy engineering as a service through clean, simple APIs.
(c) Gretel.ai
(c) Gretel.ai

Today we are excited to announce launching Beta2 for Gretel to the world on July 14th 2021. Together, we hope to automate privacy engineering as a service for anyone, and early users like you will be critical in defining what Gretel.ai becomes, so please don’t be shy with your feedback, good or bad. Most importantly, thank you for your support, and we hope you stop by to say hi in our Discord community.

In the meantime, here are some of the new features that we are most excited about! 

Making privacy engineering accessible to anyone

With Beta2, you can get started creating synthetic data and automating your privacy engineering use cases in minutes. Currently, Beta2 is completely free, but for the time being is limited to a single workload at a time and workloads are limited to a one hour duration.

The command-line interface (CLI) becomes the primary method for interfacing with Gretel’s APIs- similar to AWS, GitHub, and Terraform. There is no longer any need to write code to label, transform, or synthesize data. Instead, anyone can now edit a template YAML configuration file (or use one of our defaults) and submit tasks directly to Gretel.

It's easy to use

All you need to do is point to your datasets in CSV format, modify a template YAML configuration file, and submit your job through either our CLI or web-based Console. Labels and transformations are delivered within seconds, and synthetic models can be trained to generate completely safe and artificial datasets in minutes.

Privacy built-in by design

Beta2 implements three levels of controls to protect users' privacy in your synthetic data that are enabled by default, with support for optionally enabling DP-SGD differential privacy during model training for formal, mathematical guarantees around privacy.

Run your workloads anywhere

Security and compliance are top of mind, and some customers have regulatory and compliance requirements that the data in their cloud must remain in their cloud. With Beta2, users can choose to scale out their workloads in the Gretel cloud or deploy workers directly into their environment. When running in local mode, your data never leaves your environment.

Gretel Beta2 architecture

Our blogs and use cases are growing up

Over the past year, we have published over 30 blogs and code examples for top use cases we see in the space. We identified the most popular use-case blogs below and have promoted them to tutorials on our new docs site!

  1. Create synthetic data
  2. Label and anonymize PII
  3. Remove bias from a machine learning dataset
  4. Synthesize time-series data
  5. Smart-seeding is auto-complete for your data

Do you have an idea for a blog or a use case you want to see us support long term? Let us know!

What is going away?

Moving to an API-first approach requires us to deprecate APIs and sunset certain features from Beta1. Here is a comprehensive list of features and deprecation timelines. Do you have any questions about what is being removed? Don’t hesitate to reach out.

API and feature changes

Are there service limits?

As we continue our commitment to the developer community, we will continue to provide a full-featured freemium offering. Read full details in our docs.

How can I get started?

You can sign up for Gretel for free with a Gmail or GitHub account. When Beta2 launches, anyone will be able to install our new CLI tool or start working with our new web console to create safe data. Gretel pricing will be released early next year when we exit our beta period and move to a general availability release.

Do you have questions about Beta2 or ideas for how we can improve our services? If so, please reach out to us on Twitter @gretel_ai.