Mona Reviews

How easy you can report metrics from code and how interesting it can be to view these metrics forming a graph

Sometimes it takes time to find the relevant function I am looking for in to match my needs

Generally, the problems that we see occur sporadically, but we have no idea on what scale they do happen to see if they imply a serious issue or not. Mona helps us gain insights into such concerns by monitoring the behavior on a wide scale

Mona allows you to monitor the behavior of your machine learning models in production and alerts you when something is off. Any company with ML models in production should have a monitoring tool, and developing one in-house is a real pain. Mona is an excellent choice for this task! The integration was fast, and the support team was very attentive to any questions or requests.

I can't think of anything to dislike. In the beginning, we had some issues with duplicated insights, but it was due to a wrong configuration from our side, and they helped us resolve it quickly.

Instead of actively checking and ensuring our ML models behave as expected in production over time, I rely on Mona to find the anomalies and alert when found.

very low configuration, don't have to guess what segment my problems will arrive from or what the thresholds are.

slack messages could be improved by having an example identifier for the entity in question.

the main problem we solve with mona is quality health of our system

We don't need to define the segmentation of the data, MONA learns our data and provides fast insights. A lot of integrations out of the bots (Slack / Email).

Some of the concepts and terms are hard to understand at first sight and require some training.

Anomaly detection, production workload monitoring. MONA is our "last" line of defense in the case that all our monitoring tools fail to detect an issue.

The daily alert notifications on the chat window highlighting the issues and anomalies in the BOTs help me take corrective actions at the right time.

Nothing as such. Maybe the Archive feature could also have an option to append reasons.

Mona provides daily alerts to me about any abnormal deviations and anomalies in the BOTs which helps in taking timely corrective actions.

The best thing about Mona is that it tracks performance metrics and provides confidence in model performance.

It has great streamline team collaboration and way better than its competitor.

The third-party integration and model training in Mona needs to be developed. These need to be worked properly and products should be much better and I want improvement here.

The problem of model training and model performance needs to be developed and so here benefits need to be taken here so that it will be much more productive and business performed should be better.

There's two thing that I like the most. First, is how the company listen and cares about our needs from their platform, and try to fullfill our needs, and second is how Mona Labs are flexible to us change the config how we like. For now I'm satisfied by their product.

One thing that I dislike is the lack of extraction options from data on Dashboard. Not big deal, we can extract outside Mona, but If I would have options to extract data with the options that I would like, it would be lesser than problematic for me. For example, string columns tends to break tables in CSV, but not on JSON lines.

Suddenly data drifts. The benefits are pretty obvious to us, because we can monitoring our product on real time catching any strange pattern happening before become a real issue

Mona can be configured in many different ways and has a very sophisticated UI and configuration language. It allows you to segment the data in many different ways and have alarms and insights on all these segments independently. When something goes off in many places at once, Mona typically recognizes that and reports only one of the alarms, classifying the others as "related".

I am also very happy about our interaction with the Mona team. They have even developed some functionalities specifically for us, like seeing insights on a map.

Getting up to speed can be complicated, the sophistication of the platform makes it intimidating for some users and requires a bit of a learning curve. The documentation is good and the team very supportive, but it does require a bit of use for people to become comfortable with it.

Monitoring our machine learning models in production, by making sure that the data and the answers from the model are stable and not shifting. The configurability of the platform allows you to give whatever meaning you need to these terms.

Mona helps us monitor our models and find outlier behaviors.

The way we configured it all happens automatically and we receive alerts when something bad happens.

It is not something that I dislike, but I think our data could be better leveraged by Mona to give us some easier Data Exploration and BI tools to analyze our data.

The way we use it today, I still need a BI platform to do some analysis besides Mona.

We have hundreds of models running in production and we need to monitor them closely.

We could write our own code to monitor and detect outliers, but that would drain resources from our team that can be better used to do other work.

That way Mona is letting us focus on what is critical for our business and saving us some money.

Productionizing, monitoring the performance, improvements, investigation of underperformance, optimization, all these has been achieved using Mona, which makes the life of Data scientists easy!

Evaluating underperformance and alerting the person is okay, but many times the alert goes un-noticed. I wish it could be a message rather than alerts.

Evaluating models, detection of underperformance has always been a headache when it moves to production. Model requires a constant retraining intervals in specific time, in which Mona helps us to do. Optimization and Improvements were another plus point