Header banner
Revain logoHome Page
superwise logo

Superwise Review

3

·

Very good

Revainrating 4.5 out of 5  
Rating 
4.3
Artificial Intelligence, AI and Machine Learning Operationalization

View on AmazonView on ЯM

Description of Superwise

As more businesses rely on AI models to boost their impact and their bottom-line, the need for managing, monitoring and optimizing the real-life behaviour of these models grows. Superwise.ai is the company that monitors and assures the health of AI models in production. Already used by top-tier organizations, Superwise.ai monitors millions of predictions daily to eliminate the risks derived by these models’ black-box nature: bad decisions, unwanted bias, and compliance issues. Their AI assurance solution acts as the one source of truth for all the stakeholders, and empowers data science and operational teams with the right insights to scale their use of AI by becoming more independent, agile, and gain confidence in their models’ operations. Implemented use cases include Customer Lifetime Value (CLV) predictions, fraud detection, lead scoring, underwriting, credit risk, and more. Recognized for its innovative technology and approach, Gartner recently named superwise as a 2020 Cool Vendor in Enterprise AI Governance.

Reviews

Global ratings 3
  • 5
    1
  • 4
    2
  • 3
    0
  • 2
    0
  • 1
    0

Type of review

Revainrating 4 out of 5

Superwise - Not reliable option anymore

The fact that I can write code in Python which then gets deployed directly into my model helps me save time from having to do it manually everytime there is an update or improvement made in TensorFlow. Also, it's super easy to connect my machine learning scripts with other python libraries like Apache Airflow/SageMaker etc. There are some minor bugs where sometimes when you try to run a script using Superwise API it fails but doesn't give much explanation as to what went wrong. It could help if

Pros
  • Easy integration
  • Can deploy models without any hassle.
  • Fewer dependencies required compared than tensorboardX (TensorBoard)
  • Ability To add custom metrics within your pipeline e.g Custom Losses , Accuracy Metrics.etc
Cons
  • None so far

Revainrating 4 out of 5

A very good option for monitoring and improving Machine Learning Models

I like that it allows me to monitor my model's performance in terms of accuracy, precision, recall etc. It also gives you an idea about how your data set is performing compared to other similar models. The ease with which one can create new models using this tool is amazing! Also the documentation could be better as there are some things that aren't clear upon first use. But once you understand them, they make sense. Try out the trial version before purchasing. You will see why we decided to…

Pros
  • It is really user friendly and you can build really good models in a very short period of time
Cons
  • Sometimes the documentation is a little unclear and needs better organization

Revainrating 5 out of 5

Easy to learn and fun for quick insights into models

It's easy enough that I can use it without any prior experience whatsoever with machine learning or statistics (or even programming). The interface is simple but powerful - you get what looks like useful information about your model in an intuitive way! If there were some features which could be added such as visualisation capability this would make things easier still though they are not currently available from my perspective at least so far anyway :) Highly recommended if anyone has no idea

Pros
  • Intuitive to learn for beginners yet provides all necessary tools required once one gets used ot its structure, methods + functions; very customisable via Excel files also..not perfect mind :D But good overall!!
Cons
  • Some difficulties