Primarily, our use case for XGEN AI is to advise clients on how to use AI for conversational chatbots. A specific example of how I have used XGEN AI in my work is that we have advised clients on AI assistants that can help by acting like personal shoppers. We have also used it for hyper-personalization, looking into how to interact with the client based on the user's behavior in real time.

My experience with using XGEN AI for hyper-personalization is that it is generally very strong, but it needs to be implemented correctly. The way it really works well is that real-time behavior tracking is very fast, allowing you to give better results to your users. The recommendation engine is also very fast. The main point is that you need clean data; if you don't have clean data, it can reduce the impact and sometimes over-personalize, which can be of no use or may have negative implications as users might see repetitive items.

The best features XGEN AI offers, in my view, are its strong event tracking capabilities. It can track events, clicks, and views, and it has good product metadata. If you're looking to build a true conversational AI engine, it is the best. My assessment is that it works best when treated as a revenue engine, not just as a feature. You have to tie it to a metric such as conversation and retention to see clear ROIs.

What stands out to me most about the event tracking or conversational AI engine in XGEN AI is its conversational AI understanding. With NLPs or with most chatbots or voicebots that you would be building, the biggest struggle point is that they are very deterministic in nature, and they don't let you know what to tell and when to tell the user. With XGEN AI, I feel this is consolidated and you get a unified view.

XGEN AI has positively impacted our organization by helping us track what users are looking for. The initial release itself showed that the success rate is more than what we were getting previously. We were able to collect a lot of data, and the best part is that it can work across channels, apps, and emails, which helps us provide a unified experience to the end user.

We have seen XGEN AI recommendations lift conversion by 10 to 15 percent. We have experienced real-time behavior tracking and have started seeing some ROIs; though I'm not allowed to share the actual ROI itself, we see improvement in the overall metrics. User engagement has been very positive. We have focus groups and are collecting client feedback, and for most people that we have been able to capture feedback from, the CSAT has improved. That's the biggest thing, so overall, it's trending towards positive.

One of the improvements I would suggest for XGEN AI is the use of hybrid models and asking real quality questions to the users. Additionally, product attributes or data quality needs to be improved upon; clean historical interaction data and noise removal are necessary. This is more on our side as compared to XGEN AI itself. Better explainability is also required; when it recommends a product, we often don't know why or how. More importantly, we hope for omnichannel consistency, meaning that no matter which channel you come from, you have the same kind of experience.

I have been using XGEN AI for almost a year now.

XGEN AI is stable for us.

In terms of scalability, we are able to scale with XGEN AI. I did not see any issues, though we had to make a lot of changes to our infrastructure due to being in legacy applications, which requires many adjustments from our perspective. However, XGEN AI as a platform was stable.

Our customer support experience has been positive; we have an account manager who helps us with everything and a dedicated team. Since we were one of their largest customers at that point in time, we received ample help from them.

We did not use another solution before.

XGEN AI is deployed in my organization primarily on the cloud, specifically on AWS, because it can handle real-time data scaling and allows for faster rollout. Some parts are on-premise due to regulations and PII data requirements. However, the default setup is cloud-based.

We have started seeing a return on investment with XGEN AI. We are experiencing improvement in overall employee engagement, and we have seen some return in terms of real money. It is not at the benchmark we have set, but we are slowly reaching there.

My experience with pricing, setup cost, and licensing is that it is in line with other similar providers we have used. I would say pricing is comparable, and the licensing is based on subscription costs we are paying. The setup was a one-time cost, which was also in line with what we have paid to any other vendors.

Before choosing XGEN AI, we evaluated various other options, including the Microsoft Azure AI stack and AWS Personalize. We also attempted to build a solution in-house using Python, but none of those options fit our needs.

To get the best out of XGEN AI, you need to treat it as a revenue engine; you cannot treat it merely as a feature. I rate XGEN AI a seven overall. I rate it a seven because I have divided it into two categories: what works well, such as a strong personalization engine and measurable lift in conversion with real-time recommendations, which makes it a good fit for e-commerce catalogs. However, the things that do not work as well include its high dependency on data quality and very limited transparency in how recommendations are generated, which needs to improve. The reason I have given it a seven is it delivers value, but only when supported by clean data, which is crucial.

We use AWS as our cloud provider. I am not certain if we purchased XGEN AI through the AWS Marketplace; I think we may have bought it through there or directly through the company itself, but I do not recall the details. I have mentioned everything else that is needed for XGEN AI.

My advice for others looking into using XGEN AI is to have a strong data foundation before proceeding because garbage in, garbage out applies here. The better the data you have, the better recommendations you will get from XGEN AI. If your data is not stable, you should not expect it to work. XGEN AI is a stable platform, but you need to have a real use case to implement it; you cannot adopt it without defining your use cases. I give XGEN AI an overall rating of seven out of ten.