NiCE Cognigy - AI Agents for Enterprise Contact Centers

My main use case for Cognigy.AI Platform  is building conversational AI agents. Regarding my main use case on Cognigy.AI Platform , we're also now looking into, or maybe mainly I'm looking into the domain also for integrating MCP within the conversational AI agents.

Cognigy.AI Platform has positively impacted my organization because, on the use cases itself, the AI-driven decisions are working quite well. It is better received than the NLU solution that was in place before, so I'd say the implementation itself was a definite win.

The best features that Cognigy.AI Platform offers include the flexibility of the different nodes provided in order to develop different domains, so you are independent of APIs, or you do not have to rely on specific APIs that are only available in Cognigy.AI Platform, but you can also work with other APIs and then get it integrated. I think that is the biggest advantage.

The flexibility of Cognigy.AI Platform has helped me in my projects because it's easier to integrate it with the in-house or on-premise APIs that clients have, which makes it easier. Of course, you have to consider the security layer there, but it's almost direct access you have to the resources, considering that you also take the security into account.

Since implementing Cognigy.AI Platform, the containment rate has improved. Most of the conversations are maintained within Cognigy.AI Platform itself, and unnecessary conversations such as intents that shouldn't be processed are taken quite well and then also accordingly responded to from the AI agent.

I believe Cognigy.AI Platform could be improved in terms of versioning of flows, for example, similar to GitHub , not only on the custom extensions that you have, but also on the flows itself. If there's some versioning possible, that would be helpful, and this is one of the bigger issues.

Regarding Cognigy.AI Platform's AI capabilities, I think security-wise there are some concerns because you cannot really protect the REST endpoints, and there are some open points on the security perspective and on the governance perspective. There are options for OAuth authentication, but for the REST endpoints, for example, this was a major topic.

I have been using Cognigy.AI Platform for around a year.

Cognigy.AI Platform is stable.

Regarding the scalability of Cognigy.AI Platform, there's a limitation on the rate limit. However, with the right vendor support, there are ways to avoid these kinds of issues.

The customer support for Cognigy.AI Platform is quite good.

I previously used a different solution before Cognigy.AI Platform, specifically Dialogflow  from conversational agents from GCP . The reason we switched was that the vendor support was better with Cognigy.AI Platform than with conversational agents from GCP .

Before choosing Cognigy.AI Platform, I evaluated other options such as Dialogflow  or conversational agents from GCP. This was one of the first choices because we initially had some experience there with that product.

My advice for others looking into using Cognigy.AI Platform is to complete the training that is offered from Cognigy.AI Platform itself to get an idea of how flow building and everything works, and to get a high-level overview. On a second level, use metaprompting to build your flows because a lot of the components can already be co-developed using metaprompting.

The development itself on Cognigy.AI Platform is quite self-explanatory. Via metaprompting, you get all the resources, and it's also easily configurable. The UI is quite friendly, but far from perfect. In terms of Cognigy.AI Platform's AI capabilities regarding its accuracy and reliability of output, it's quite good. It depends on the language model you use, but in general, all functionalities work quite well. At the end, it depends on the end-to-end testing and the quality of the data you have. On a scale of one to ten, I would rate Cognigy.AI Platform an eight.

I built an agentic bot for customer support with Cognigy.AI Platform , where I had the possibility to come up with my own architecture. I constructed both an agent that was performing intent recognition. To understand the accuracy of the agent, I built a judge that was an LLM system evaluating the agent responses and intent accuracy recognition. This was useful because since the analysis section in Cognigy.AI Platform  is limited, by using another system to judge the decision, I was able to track how accurately the system was performing intent recognition.

The academy resources provided by Cognigy.AI Platform are valuable for getting started. After learning everything, there is a test that, once passed, provides a certification, which is useful as a backup document certifying that I am a developer in Cognigy.AI Platform.

In terms of building flows and working with the platform, the setup is easy to understand. There is no backend black box that could allow me to mess something up. The platform allows sharing projects with different people who can work simultaneously on the same project. I can see who is changing what and when, which is a strong feature.

Cognigy.AI Platform's natural language processing capabilities have helped streamline communication within my organization in positive ways.

Cognigy.AI Platform has enhanced my customer interactions because all clients have been positive about it and have seen the benefits. When connecting cloud systems with different sections that are organized, everybody can see what goes where and when. There is transparency that the client values.

Cognigy.AI Platform is limited from a developing perspective, especially regarding the analytics aspect that should improve. When debugging, there is no direct page for this. I have to go to the logs and see what errors appear each time I run the flow. There is significant room for improvement where the platform needs to think from a technical perspective and support different personas, both non-technical and technical.

For technical people, it would be beneficial if the platform could connect to Visual Studio Code  so that technical people can easily debug without having to go node by node and adjust things whenever errors are thrown.

The graphs in the analysis section of Cognigy.AI Platform should display metrics such as average duration of calls or number of conversations, which would be helpful if visualized internally.

I have been working with Cognigy.AI Platform for six months.

Cognigy.AI Platform is performing accurately, but to do so, I need to restrict the LLMs properly. There is a lot of back and forth in terms of system prompt adjustments. If I rely solely on the LLM, things would go downside. I have to be very careful about how I am defining and setting up the prompt engineering.

I find Cognigy.AI Platform scalable.

The technical support team takes some time to provide a response.

The initial setup of Cognigy.AI Platform is straightforward. I just have to define the agent prompt, set up the model that I want to use, and then move forward to building the actual workflow. This is similar to what ElevenLabs  does, so on that regard, they are the same.

Cognigy.AI Platform's pricing is expensive for the service that is offered.

In comparing Cognigy.AI Platform and ElevenLabs , the analysis section of ElevenLabs is different because it offers a different number of conversations, average duration, and total cost, all set up internally.

Pros for Cognigy.AI Platform include the easy setup and the possibility to build everything from scratch instead of predefining it as is done in ElevenLabs.

Pros for ElevenLabs include the agentic bot that I can speak to directly, which is not just doing everything within a flow. This is a feature that Cognigy.AI Platform is lacking. The language configuration within ElevenLabs gives a lot of freedom to target different markets.

The main benefits that Cognigy.AI Platform brings to the table are easy setup and the academy. Clients now rely more on Cognigy.AI Platform because ElevenLabs is relatively new to the market. If I can provide Cognigy.AI Platform to them, they will be very happy because they already know the platform well.

Given my experience with Cognigy.AI Platform and ElevenLabs, my advice for users looking into Cognigy.AI Platform who would like to start working with it is to check all the videos and get the certification first. It would be useful to complete this within a week because it is absolutely achievable. Then, do not be afraid to start being creative and setting up things so you can understand how Cognigy.AI Platform thinks, how it works, and what kind of freedom is available when using it.

I would rate this product a 7 out of 10.

Over the past year, my main use case for Cognigy.AI Platform  has been to build both voicebots and chatbots for B2C use cases across several industries. Projects have included a voice assistant for an automotive dealership that automates appointment scheduling, handling identity verification, customer data lookup, and escalation without using staff involvement, as well as projects related to an internal employee-facing chatbot for a large corporate to handle ID and document processes, automating this entire process with self-service flows such as identity verification, document request, and appointment booking, as well as developing a multilingual, agentic hotel chatbot answering guest questions using both German and English.

The project for building an assistant for an automotive dealership stands out as the most interesting and challenging because we had to integrate multiple databases and use API endpoints, which made it complex because of the way the endpoints were also built. Understanding users' requests and intents regarding when they want their input was very challenging as well.

One of the best features Cognigy.AI Platform  offers is being agentic because when I previously started working with Cognigy.AI Platform, it mainly was rule-based using NLUs. Currently, having agentic capabilities and using the tools is very useful, as well as having Playbooks because they really automate the way that we test. We used to do manual testing, and with more engagement with Cognigy.AI Platform, we have started really using Playbooks, and they are a very helpful feature.

We have mainly used Playbooks for testing on Cognigy.AI Platform, so we have created our use scenario and what we expect to have from our flows. Instead of going through manual testing where we have to type everything one by one, we run a Playbook, and it checks whether the feature is completely developed as it should be or if there are any edge cases that are happening. This simplifies the way we test and really helps to stay on track with the use case that we are creating or implementing.

Cognigy.AI Platform has really helped us positively in our organization with our customer projects because the projects that I have worked on have been customer-related. We also use it internally for automating different tasks, such as ID card registration. Instead of employees going through the entire process and communicating with different colleagues to just get access for a certain building, having to do it via Cognigy.AI Platform is much easier and more simplified.

One of the biggest challenges I continue facing while using Cognigy.AI Platform is the fact that the platform lacks proper version control, and this makes it difficult when I am working with multiple team members and it is very hard to track the changes that I do or the changes that my peers are doing.

One improvement for Cognigy.AI Platform would be creating a version control, which would really help developers. This way, we can keep track of what we are doing and how we are doing it. Additionally, if we could make an integration with IDEs just like Visual Studio Code , it would make easier implementation for the coding part and just to have an integration between an IDE  and the local platform. I think that would be an interesting feature.

I have not had the chance to have a lot of communication regarding customer support because it is handled by another level of support for us, but it has been quite useful. Every moment that we have had the need for customer support from Cognigy.AI Platform, they have been there for us, solving our issues and guiding us.

If I take it from the customer side point of view, I think we have seen a return on investment. Automating customer service really reduces the number of employees needed during peak hours, which is an achievement, especially if you are dealing with multiple calls per day. This would reduce a lot of the work for the employees.

Cognigy.AI Platform has really helped us positively in our organization with our customer projects because the projects that I have worked on have been customer-related. We also use it internally for automating different tasks, such as ID card registration. Instead of employees going through the entire process and communicating with different colleagues to just get access for a certain building, having to do it via Cognigy.AI Platform is much easier and more simplified.

I would suggest getting into Cognigy.AI Platform. The more you start working with it, the more you can see that you can do a lot of great things. You learn by doing and by being curious about what Cognigy.AI Platform can offer.

Regarding Cognigy.AI Platform's AI capabilities, I would say its governance and security are good so far, but this is not one of the scopes I focus on.

When it comes to the accuracy and reliability of output from Cognigy.AI Platform, the information is accurate. We just need to always be careful with the guardrails, mainly when prompting. I would rate this review an eight out of ten.