Showing results for 
Search instead for 
Did you mean: 

DataRobot vs Integrated Enterprise Application AI

Image Sensor

Why would an organization that already has an enterprise application such as SAP or Salesforce want to use DataRobot over the enterprise application's integrated AI platform or extension such as Einstein in the case of Salesforce?

3-5 reasons please.

2 Replies
DC Motor

Hi SD -

That is a good question. Another question might be, if an organization already has SAP, why would they also choose to go with Salesforce with CRM? Ultimately this comes down to the breadth and depth of capabilities offered. What makes DataRobot such a good choice?

  1. DataRobot is not simply an extension - but a dedicated platform built from the ground up to address machine learning needs across the entire spectrum, from data prep, to model building, to model hosting and monitoring.  We are not a partial solution; these are not components coming some day - this is a fully complete world class solution that is available to you right now.
  2. We are creators and leaders in the automated ML space - we've been at this a long time honing our craft!
  3. We address the needs of both business analysts as well as sophisticated data scientists.  Analysts can leverage the platform to build and deploy trusted models despite not having coding or statistical experience.  Data scientists can augment their capabilities with our platform to accelerate their productivity and put their output into overdrive.  Many platforms try to cater to one group at the sacrifice of another; we serve as a powerful platform for both.
  4. We offer many ways to integrate with systems, and a large variety of model deployment and usage options.  Our integration points are platform agnostic, and we are designed with flexibility around a large number of integration options rather than forcing any workflow.  Eg. want to start your project from a csv file on your local machine?  Training file in AWS S3?  Data inside a Snowflake data warehouse?  How would you like to deploy your model - we can host it for both batch or real time processes.  We can also provide a docker container to run it where you'd like; or as a java binary file to run it offline; as well as the capability to still monitor both, among even more options!
  5. With DataRobot, you get access to a talent pool of hundreds of experienced data scientists globally as well as data engineers to help you integrate with the platform and your models.  You may have seen that many data science projects fail or never make it to production - we're here to help and invested in you getting your 1st and 100th model deployed and used, so you start recognizing real value out of it.  


DataRobot Employee
DataRobot Employee


In summary it comes down to functionality and flexibility, and providing your analytics teams with a more complete toolset. I would suggest that the following are often drivers for this decision:

  1. A company using the inbuilt capabilities realises that data to help solve the business problem they have resides in other systems. This will often lead to a decision to invest in a separate analytics solution that is not constrained by using data only in the source system. This is especially true where investment has already been made in enterprise data stores that bring together data from multiple systems.
  2. In-built integrated components are based on the expectation that you are going to use them in a specific way, whereas most businesses have unique idiosyncrasies that make using these components frustrating for users and time consuming for your developers. 
  3. Model/algorithm choice - the integrated components have already chosen the algorithm they are going to use, and are difficult to customise (if customisable at all). DataRobot provides access to a wide range of algorithm choices that lead to a better outcome.
  4. Business adoption - business users won't use the inbuilt solutions for a variety of reasons, largely due to them being a 'black box'. With DataRobot they can see exactly what is driving the predictions, both at an overall model level and at an individual prediction level.

Hope this helps.