In a space crowded with solutions coming from the open source world and commercial vendors, it’s sometimes difficult to understand precisely who is who and who does what. In this context, one of the most frequently asked questions to our team is “What makes Prevision.io different?”

 

While I won’t explicitly cite the differences between Prevision.io and every solution on the market (would be a two days reading exercise and would be deprecated in a week after publishing), I will highlight the key differences between Prevision.io and the various solution types on the market today:

  • Cloud Vendors
  • Data Science & Machine Learning vendors
  • MLOps (deployment & monitoring) vendors
  • Open sources modelling or “AutoML” frameworks

 

Prevision.io

In order to move forward, you first need to have a clear understanding of who is Prevision.io and what do we do:

Founded in 2016 by a team of renowned data scientists, Prevision.io brings powerful AI management capabilities to data science users so more AI projects make it into production and stay in production. Prevision.io’s purpose-built AI Management Platform was designed by data scientists for data scientists and developers to scale their value, domain expertise, and impact.  From banking and financial services to healthcare and retail, data scientists too often lack the tools to create efficient data models. Now with Prevision.io, a member of the Google Cloud Partner Advantage program, data scientists and analysts have the tools they need in one place to build, deploy, monitor, and manage data models across a variety of industries.

What makes us unique?

 

Cloud vendors

Cloud vendors, typically the major ones like Google, Azure, or AWS all have offerings aimed at helping Data Scientists or Machine Learning Engineers to actually build and deploy Machine Learning based models. However:

  • We have a quicker set-up phase thanks to our SaaS offering. There is no need to master IAM for people can be up and running on Prevision.io
  • We have a more coherent UX, within a unified and easy to use product that covers the whole model lifecycle (creation, deployment, monitoring, alerting, retraining) whereas they often have one specific tool for each need, requiring your team to know each of them and how they work together
  • We are aimed at both Data Scientists and Citizen Data Scientists, offering either a complete UI for clickers or notebooks with R & Python SDK for coders
  • We take care of the whole underlying infrastructure. No need to select computing resources, we do it automatically based on the usage you have
  • We offer a very competitive pricing compared to these behemoths coupled with more predictive power (see our blog post comparaison versus Vertex AI AutoML) ultimately yielding to better return on investment
  • We support the deployment of AutoML based models or your already trained models in one click without the need to write a single line of code
  • We offer model monitoring capabilities complete with an alerting system to make sure that every model and applications work as they should without human intervention

Data Science & Machine Learning vendors

Data Science & Machine Learning vendors have a value proposal that has some overlap with Prevision.io. However:

  • We offer a unique SaaS offering with Pay As You Go billing, without any recurring subscription fees which is unique on the market. You only pay for what you use allowing us to truly democratize access to AI
  • We offer a more streamlined experience for our users. No sci-fi looking application nor hundreds of clicks required to make stuff work
  • We offer stronger deployment, monitoring and alerting capabilities than most vendors (especially the ones more focused on experiment and modelling)
  • We offer integrated development environments as packaged notebooks (R & Python), allowing coders to work alongside clickers and business users within a unified UI to enhance collaboration
  • We offer top-notch modelling capabilities thanks to our state-of-the-art AutoML platform, allowing our customers to solve their problem based on tabular, textual or images data with minimal time and resources

MLOps (deployment & monitoring) vendors

MLOps vendors specialize most of the time only in the deployment and monitoring of Machine Learning models (and to a lesser extent Machine Learning applications).  They are fairly new on the market and appeal to organizations that have already built their first iterations of models. However:

  • We also address companies that want to build new models and/or keep track of all experiments without the need to buy, implement, and maintain another dedicated solution
  • We offer strong modelling capabilities thanks to AutoML, thus allowing our users to quickly retrain a new version of a deployed model should the monitoring report any instability (this task can even be automated within Prevision.io)
  • We offer development environments that allows our users to build advanced analytics on monitoring, but also on data and/or on the model performance that most of the MLOps vendors can’t achieve because of their lack of model insight
  • We offer deployment capabilities for your models without the need to manage and maintain the underlying infrastructure, APIs or autoscaling
  • We offer advanced monitoring analytics for all deployed models, without the need to edit your code

Open sources modelling or “AutoML” frameworks

To a lesser extent, we can compare Prevision.io to some modelling or “AutoML” frameworks available for free in the open source world. However:

  • We provide a complete offering, covering the whole lifecycle of Machine Learning model creation, deployment, monitoring, alerting and retraining within a coherent UX which is way beyond the scope of the underlying framework. This removes the need for users to be skilled across each discipline of the cycle
  • We provide a cloud based solution ready to go in a couple of minutes without any installation or provisioning of compute resources
  • We offer a friction free offering thanks to our Pay As You Go model which removes licensing burdens
  • We offer a friendly UI alongside a full code friendly environment to our users
  • We offer dedicated support and training to our customer that is not available for open source frameworks

 

In the interest of brevity, I only addressed key features and measurable differences on value proposals. If you have any more questions or comments, feel free to reach out or better, register for free now at Prevision.io and draw your own conclusions.

Florian Laroumagne

About the author

Florian Laroumagne

Senior Data Scientist & Co-founder

Engineer in computer science and applied mathematics, Florian specialized in the field of Business Intelligence then Data Science. He is certified by ENSAE and obtained 2nd place in the Best Data Scientist of France competition in 2018. Florian is the co-founder of Prevision.io, a startup specializing in the automation of Machine Learning. He is now leading several applied predictive modeling projects for various clients.