We primarily use Fiddler AI to monitor our AI models in production and catch issues like drift or unexpected prediction changes early. For example, if a client-facing model starts behaving differently after new data comes in, we use Fiddler AI to trace the problem, understand why it happened, and decide whether the model needs adjustment or retraining.

As a web agency, we have been using Fiddler AI for about eight months now, and it has proven to be excellent. Fiddler AI fits into our workflow as an early warning and debugging layer for production models. It helps us spend very little time manually investigating issues and allows us to focus more time on fixing the actual root cause, which makes our model operations much smoother. Over time, it has become part of our regular workflow whenever we need to check the model health or investigate unexpected behavior. Instead of starting from scratch, we can quickly narrow down to the issue and move faster on resolution with the help of Fiddler AI. For us, it is not just a monitoring tool but rather part of how we keep our AI systems reliable every day.

The two features we rely on most are drift detection and explainability. Drift detection helps us spot issues early, and explainability helps us understand the root cause quickly so that we can take action faster. Together, those features make it easier to monitor and debug models in production.

The alerting feature is also helpful with Fiddler AI because it lets us respond quickly when something changes. The interface makes it easy to move from an alert to a specific model or data slice that needs attention. This saves a lot of time and keeps the workflow smooth.

What helps most with Fiddler AI is the combination of drift detection and explainability. We can first see when the performance or prediction patterns have changed, and then drill into the specific features or slices of data causing the issue. That makes it much easier to figure out whether the problem is data drift, a bad input pattern, or something in the model logic. The most valuable aspect is that Fiddler AI shows both the change and the reason behind it, so we can trace the issues from an alert into the exact data slice or feature that caused the problem.

The main thing I would improve with Fiddler AI is onboarding and ease of use for different stakeholders. The platform itself is very powerful, but making some workflows more guided and adding more ready-made dashboards would make it even better for a team in our situation. It is strong overall, but simpler onboarding, better dashboards, and more automation would make it even easier to use.

I have been working in ADP for about two years and six months to be precise.

Fiddler AI has been stable for us overall, and we have not run into any downtime or reliability issues. It has been dependable for day-to-day monitoring and investigations, and I commend the team managing it.

Fiddler AI has scaled very well for us so far. It has been able to handle our workload even better as we add more models because it gives us a centralized way to monitor and debug everything. From what we have seen, it scales very effectively, and as our usage grows, it still feels manageable and does not become a bottleneck for the team.

We have interacted with Fiddler AI's support team very rarely, but the experience has been very good. They are very responsive, professional, and helpful when we needed clarification or guidance. The support overall was excellent.

We were using a more manual approach before Fiddler AI, along with some basic monitoring tools. We moved to Fiddler AI because it gives us much greater visibility into model behavior and makes debugging much faster.

The documentation, integrations, and everything else look perfect. The only suggestion would be if they could add simpler onboarding, better dashboards, and more automation, as it would be even better.

I am satisfied with Fiddler AI, and I have discussed everything with the relevant parties.

We have clearly seen the return on investment mainly through the time saved and more efficient use case of our eight-member team. We spend much less time manually investigating model issues, which has become a very significant advantage. This means fewer hours lost on debugging and faster incident response. We have not had to add extra headcount just to manage model monitoring, so the platform has helped us stay efficient as we scale.

Pricing for Fiddler AI feels appropriate for an enterprise-level product, so it is not the cheapest option, but the value is there for a production AI use case. The setup cost was manageable since it was a hosted platform, and licensing was straightforward for a team-based deployment.

We did look into a few other options before choosing Fiddler AI, mostly general ML monitoring and observability tools, as well as our internal manual approach. Fiddler AI stood out because it was a strong option on explainability, drift detection, and making production debugging easier.

My advice to others looking into using Fiddler AI would be straightforward. If you need serious production monitoring and faster model debugging, go with Fiddler AI. It is especially valuable if your team wants better visibility into drift, explainability, and incident response. I recommend starting with a clear workflow and making sure your team is aligned on what to monitor so that you get the value from it very quickly. I have given this review an overall rating of 8.

I am currently working with D Innovation Private Limited for a client called Casia in a Senior Data Engineer profile, where I have been working for the last 1.6 years. I have a total of eight years of experience working at Infosys as well.

I use data engineering tools for Power BI and Fiddler. For analysis, I have to create a model to analyze the data. I utilize our data for this purpose.

In day-to-day work, if I need to support or create any language-based model with features, I find it very easy to use. I create modular solutions that convert from one language to another. Fiddler AI is the best way to handle this right now.

I am utilizing Fiddler AI for AI agents with multiple Matillion tools and Kubernetes. I mention that we have AI agents for these purposes. The pricing is good, at approximately 0.002 dollars per developer, which is very cost-effective. Additionally, the latency is less than 100 microseconds, which is excellent for handling Fiddler AI.

I have been using Fiddler AI in multiple ways for approximately two to three years.

Fiddler and Fiddler AI serve different purposes. I utilize Fiddler for creating Power BI reports, and it is very good visually. There are multiple options for color choices and data design, allowing me to explain data effectively. Right now, I utilize Fiddler AI for machine learning purposes.

In Power BI projects, I use it for data enhancement and data quality checking for schema validation. I use it for data analysis as well. I connect it directly to my reports through Fiddler API and write Python code for analysis. I integrate this into my Power BI projects.

All features in Fiddler AI are good, offering excellent connectivity options. The REST API is good, supporting multiple languages with connected speech and responsive answers for any AI tools. Data performance is very good, and it is very easy to use. From a learning perspective, it provides responsive answers for data manipulation and other utilization needs.

To improve things, I need to connect it for handling LLM prompts and data management. This is mainly for understanding queries better. Snowflake Cortex is similar, but it does not give very good responses and sometimes provides poor experiences. Fiddler AI provides a better way.

I should not give too much data to any AI agent because when an agent reads and analyzes data itself, this is the wrong approach. Giving full data access to any organization is not a good practice. I need to analyze any tool first before proceeding. After that analysis, I can move forward.

I have been using this solution for approximately six to seven years.

Fiddler AI is stable and performs well.

Scalability is very good. The solution handles 7 to 15x more data, which is a very good opportunity.

Customer support is very good. We handle data and data breaks well, and Google Cloud support is available anywhere, which is excellent. Fiddler AI supports healthcare, AI defense, intelligence, healthcare data, and insurance data.

I did not use a previous solution and have just now tried Fiddler AI.

The initial setup is fine with no issues.

The implementation team was fine.

A significant amount of time is saved, which is very good for us. Both time and money are saved. If I needed to utilize a task with 10 people, I can now handle data analysis with only five people, which is very good for our organization.

Licensing cost is very low, which is very good. The usage cost is reduced by 70 to 80x compared to other solutions, and the latency is less than 100 microseconds, which also provides a very good value.

I have definitely considered other solutions.

I will need to check on further recommendations after some time. I rate this product at 8.5 out of 10.

My main use case and interest in Fiddler AI is detecting and analyzing AI models. If we can analyze the AI model, it's a significant achievement for us, and we can easily implement AI features.

A specific example of how I tried to analyze and detect an AI model using Fiddler AI involves my work in the QSR domain. Imagine there is a food delivery app. Fiddler AI can be used to recommend restaurants, recipes, and nearby or favorite restaurants.

I found Fiddler AI to be the best solution in the market to analyze the QSR domain. As a QSR domain expert, we can use Fiddler AI to give personalized recommendations. If we have an application for restaurants or QSR, we can personalize the menu, homepage, and cart to a specific user. If I am an end user using an application and receiving menu choices based on my previous recommendations, that would be an eye-catching moment that we can easily achieve using Fiddler AI.

The best features that Fiddler AI offers include monitoring and managing the model, which is the best feature.

When I say monitoring and managing the model, the aspects I find most valuable are tracking performance, detecting drift, and alerting when something goes wrong. Fiddler AI helps me analyze behavior and a bunch of data, since AI is nothing but a set of data. I can track user journey and analyze user behavior.

Fiddler AI has positively impacted my organization through my exploration, even though I haven't implemented it in my company. I am an AI enthusiast, so I try to explore solutions that I can recommend to my company. Based on my exploration, I think Fiddler AI could have a positive impact if my company decided to use it. It has the potential to help my application personalize recommendations according to the user, which is the main trending feature of any QSR app available in the market.

If we can monitor the models, we are monitoring the behavior of the customer. If we can offer what the customer wants, it will directly and indirectly increase sales and automatically generate more revenue for us.

I would not say Fiddler AI needs improvements. Whatever Fiddler AI offers is great to have.

I have been exploring Fiddler AI. I haven't used it particularly, but I explored it out of curiosity to learn AI. I found Fiddler AI to monitor and manage machine learning and LLM models.

Fiddler AI is stable from what I explored.

Based on my exploration, I would rate Fiddler AI's scalability as eight on a scale of one to ten. It depends if my user base increases, which part I haven't tested, so I don't want to comment on that.

The customer support for Fiddler AI, based on my interactions, is looking good. I'm interacting with you, and I would say I had a great experience connecting with you.

Positive

I haven't previously used an AI solution for model monitoring or AI explainability. I implemented an in-house personalization solution providing recommendations. There was a platform we used for personalization, but I can't give the name of the company due to confidentiality.

I haven't reviewed the pricing, setup cost, and licensing while exploring Fiddler AI because it's not my area of focus. I recommended it to my organization, and they have a specific team or finance team who will review everything, and if the budget allows, they will definitely pursue it.

I definitely anticipate specific benefits and money growth if my organization adopts Fiddler AI. It will help us quickly adopt and implement solutions.

I did not evaluate any other AI platforms before considering Fiddler AI. In the market, there are already multiple personalized platforms. We explored Sitecore as well.

If I get the opportunity to use Fiddler AI in my organization, I would keep it in my organization first, then go with the public and hybrid part by using the cloud. If I were to use a hybrid cloud, I would prefer to use either AWS or Azure, depending on the budget my company provides.

My advice to others looking into using Fiddler AI is simply to monitor and detect the models and user behavior. Fiddler AI is the best solution in the market, and I would give it my strongest recommendation. My overall review rating for Fiddler AI is nine out of ten.