Radware AI Security Suite Reviews

My main use case for Radware AI Security Suite is monitoring AI usage of the humans in the company and it creates rules that I use to monitor employees' AI activity and create rules around it.

Monitoring employee AI activity and setting those rules is especially important for my company because as a security specialist, I want to know what the search agents of the employees in my company are and if their prompts can be used for harmful purposes for the company.

My understanding of potential risks, such as shadow AI or tool misuse, evolved as I heard many issues and risks according to AI usage and the prompts that can leak information of the company. I told this to my CIO, and then we started with the purchase of Radware agentic AI.

Among the concerns of data leakage and tool misuse, our organization prioritized the issue of data leakage most when considering Radware agentic AI protection.

The specific LLM-related risks that were most critical for my organization to address with Radware LLM Firewall include prompt injection and sensitive data leakage. Prompt injection is the primary issue in our organization, along with XSS.

The specific AI platforms my organization prioritized connecting first with Radware solution include only Copilot, but I am also using Gemini.

Radware AI Security Suite has positively impacted my organization because I create the rules for blocking special things that can leak outside from the organization, and that is what is good in the product.

I can share that I observed someone from the FinOps team sharing with Copilot any file with an ID, and Copilot blocked it with this product.

Upon activating Radware agentic AI protection, I identified approximately 300 agents of Copilot in my environment.

I did not discover any unexpected insights about my internal network topology with Radware agentic AI protection.

The specific types of suspicious prompts, sensitive data exposures, or unsafe model responses that Radware LLM Firewall detected or blocked since deployment include blocking all issues with ID numbers, medical information, and bank accounts.

The best features Radware AI Security Suite offers include the option of the logs and the ability to see all the conversation from the end user with the AI.

The logs help me investigate or respond when something looks risky because with the logs and the rules that I create, it helps me to detect issues.

Regarding governance and security, the rules and the dashboards that I create are critical.

The specific technical demonstrations provided by Radware during my evaluation instilled confidence in its ability to protect my AI agent environment because Radware agentic AI caught my company using Python execution and the prompt injection.

Everything that you mentioned could be improved in Radware AI Security Suite.

I have been using Radware AI Security Suite for approximately one year and I am very satisfied that I purchased this product.

Radware AI Security Suite is stable. The speed of the product has not had too much effect on performance, but it is also good.

With only Copilot, Radware AI Security Suite handles my growing number of Copilot users and activity well.

I received false positive alerts from Radware, and the company helped me to tune it.

The customer support and technical support are excellent, and they are available to me anytime that I need.

I would rate the customer support as 10 out of 10.

I have not previously used any different solution before Radware AI Security Suite.

When integrating Radware agentic AI protection, the factor that influenced my decision was that the integration was performed via API in Azure.

I am a single person in charge of security, and it took me one week to set up the initial environment.

My approach to managing agent access control changed after implementing Radware solution as I created a rule in my firewall and implemented the product in Azure, and the integration was performed via the API in the cloud.

My organization's latency requirements for AI agents influenced the choice between out-of-path monitoring and in-path enforcement with Radware agentic AI protection as I wrote the requirements with the team of Radware.

I have seen a return on investment from Radware AI Security Suite. I am very satisfied with the product.

The pricing is not my concern, but the product is not too expensive and what the product offers according to the price is worth it.

I would rate Radware AI Security Suite overall as a 10 out of 10. I give it a 10 out of 10 because I like the product and all the features in the product.

I think the DDoS protection from Radware may have prompted my organization to explore solutions like Radware agentic AI protection at this time, but I am waiting for the approval of the CISO.

My advice to others looking into using Radware AI Security Suite is that I very much recommend other companies use Radware AI Security Suite and Radware agentic AI.

The most critical LLM-related risks we address with Radware AI Security Suite are proper control of prompting and gateway management. That's the main one.

We use all of them, such as Copilot, Bedrock, or anything else, and check them out, but it depends on the pricing.

Our approach to managing agent access control after implementing Radware AI Security Suite is that we're using the add-on under all security policies by the company with monitoring, of course. The observability side and all things include the usual basic monitoring, so nothing special to highlight for monitoring that.

The most valuable features in Radware AI Security Suite are debug capabilities and helping define malicious activity, but it still requires proper control. It can block whenever needed.

Our understanding of potential risks changes with Radware AI Security Suite's Agentic AI protection, since that's a new additional layer of maintenance support and all things. So that's providing some benefits, but security and additional support maintenance are questions.

Radware AI Security Suite's LLM Firewall affects response time and user experience for our internal applications in the same ways. Nothing is affected or improved in that area.

Upon activating Radware AI Security Suite's Agentic AI protection, the number of agents identified in our environment is the same as our prior estimates of active agents. Nothing is zero or estimated or underestimated, so everything regarding the architecture design of implementation is accurate.

I see room for improvement for Radware AI Security Suite. There are a lot of areas. It's a growing market with a lot of functionality that can be adopted, including observability, visibility, gateway functions, and pass-through things.

The main areas I see for improvement are performance and the security part. They need to be more transparent in that area and improve it.

In the future, I would like to see additional features that make it closer to ten out of ten. As I mentioned, more clear visibility, observability, and control of AI security would be beneficial.

We have been familiar with Radware AI Security Suite for a while. For AI security, it was released about six months ago, and we have been using Radware AI Security Suite for six months.

When it comes to rating technical support from one to ten, technical support is awesome, but the technical side needs a better product. Technical support is awesome, and they take care of issues, but technical issues sometimes get stuck by our product and some specific policies. Specific things are required to be done. From my point of view, it needs to be clear and visible, as I mentioned. When you don't know what to do, you can't fix it, and that affects you, so you need to make a support call. The support is responsive and takes action, proceeding in one day, but that's time lost and it's affecting customers, and after that, you have to calculate your ROI.

Support is great. I would rate it nine out of ten, maybe eight out of ten, not ten out of ten. Not one hundred percent, but it's around eighty to ninety percent.

The estimated engineering effort required to connect the initial environment is about ten percent of the effort, not more.

Regarding return on investment, I'm not sure we calculate it properly. For us, it's new functionality and we are not calculating it for this product because it's security, and it's always budgeted around that. So it's not benefiting or providing anything around internal rate of return or any kind of financial metric. It's just security and trust.

Regarding price, I think we are good with pricing because, as I mentioned, we are part of the portfolio and we have a discount. However, generally compared to the market, it's not cheap. To be on top, they need to provide fair pricing for all products, not only for AI.

Among the concerns of data leakage and misuse, we are in trust with Radware because, as I mentioned, we are in the same portfolio for another business. I think that's one of the questions when something is running in our infrastructure. Always the security question is in that field.

Radware AI Security Suite's demonstrations instill confidence in us, since we had a great demo and all things regarding that.

Our organization's specific latency requirements for AI agents influence the choice because those are usual requirements of five hundred milliseconds or one second. Those are usual requirements everywhere. So if it's slower, that requires improvement.

I purchased Radware products directly.

Overall, I rate Radware AI Security Suite six or seven out of ten. I would tell anyone researching Radware AI Security Suite on PeerSpot that there is significant potential in this solution, but performance optimization and greater transparency around security capabilities would strengthen its offering considerably.