AI For Retail

The main use case for Kore.ai  was building a chatbot for one of our insurance clients who required a chatbot to allow agents to directly ask about documents uploaded for verification. The chatbot checks if documents exist at a particular link that contains the personal documents of clients and then sends the status back to users or agents. If the document is present, agents may request to read it and check if certain conditions are fulfilled in the document. This is essentially a document validation agent created using Kore.ai  as the front end, coupled with Teams channels for the agents.

Kore.ai helped our team specifically in building the document validation chatbot by serving as the front end of the entire development. Using Kore.ai, we connected with the Teams channel and leveraged features such as API connections and API calls to easily retrieve the status of documents.

The best features Kore.ai offers include a very user-friendly interface, making it easy for agents who are not comfortable with technology to navigate it. This is the main aspect I would highlight.

The user interface of Kore.ai is so user-friendly because it is a no-code platform that provides drag-and-drop tools for building chatbots or agents without needing to write code. Kore.ai also offers excellent connectors with different services, and in our case, we built a microservice that was called directly from Kore.ai through API calls. Additionally, it offers popular application integrations such as SAP, ServiceNow , and Salesforce . These features have significantly improved the customer experience and made it scalable enough to manage millions of interactions, which suits large enterprises.

Kore.ai has positively impacted our organization by reducing error rates from around fifteen to twenty percent to just two to three percent after implementation. This significant reduction in errors means that fraud detection is not negatively impacted. It has also helped reduce the time agents spend on document validation, thereby increasing efficiency and accuracy. These are the main two metrics that have contributed to our success.

In terms of time saved, the accuracy has dramatically increased, with error rates decreasing from fifteen to twenty percent down to two to three percent. We have saved almost two hours of daily work per agent. With more than twenty agents working on this document validation task, we are saving a total of around forty hours per day.

To improve Kore.ai, I suggest focusing on more agentic automation, such as offering MCP kind of features with an orchestration layer for use cases. This would allow us to implement business logic in the orchestration layer. With the ability to build our own MCP servers and plug-and-play with the MCP client, we could have more scalable options for deploying multiple agents on one platform, enabling them to work simultaneously across different use cases.

I have been using Kore.ai for around eight or more months in my previous company, where I worked on a proof of concept for that particular engagement.

Kore.ai is stable.

Kore.ai's scalability is good, and it is indeed scalable.

The customer support for Kore.ai is really good.

Before Kore.ai, we used InteliX, which was one of the vendor products we explored but switched to Kore.ai due to its superior connectors and more affordable licensing compared to InteliX. Additionally, the user experience in Kore.ai is much more user-friendly, unlike the complex user interface of InteliX.

My experience with pricing, setup cost, and licensing has been good. Although I was not directly involved in the pricing discussions, the setup costs and licensing were straightforward, and we received excellent support from the Kore.ai product team. Their training sessions were effective, and we also achieved certifications through mini-projects alongside the training, making the transition and onboarding process quite smooth.

We have seen a return on investment, particularly in full-time equivalent count saved.

We did not evaluate any other options besides InteliX, but we also tried Copilot. Unfortunately, Copilot did not deliver good accuracy during the proof of concept, and its licensing costs were higher than those of Kore.ai, which is why we ultimately chose Kore.ai.

Kore.ai has impressive features, such as effective connectors and a very good user interface. It is a no-code platform where drag-and-drop functionality is sufficient to build logic without needing to write code. It is easy to deploy agents and much more scalable than other tools I have explored so far.

My advice for others considering Kore.ai is that it is really easy to use for beginners. People will quickly learn how to use it and deploy their own use cases. Kore.ai is a good product, and I would rate it an eight out of ten.

I have been using Kore.ai  for several months for certification purposes.

My main use case for Kore.ai  is a chatbot that I use for customers in the informatics area.

I do not use any other tasks or features in my chatbot beyond this primary use case.

The best features Kore.ai offers are that everything is centralized in an online platform and it is easy to use to build a production-ready chatbot.

The platform is easy to use because everything is very direct for implementation and all is on one page, which means it does not require extensive technical knowledge.

Kore.ai has positively impacted my organization. Although we did not use it for many use cases, we used it for one use case where I noticed significant improvements. It saved a lot of time in implementation. The proof of concept was very fast to implement and easy to develop.

I do not see any need for improvement in Kore.ai, as I found no challenges or missing features.

However, there could be more documentation in the latest versioning of the platform. I found that the previous documentation was outdated.

I have been working in my current field for four years.

In my experience, Kore.ai is not stable.

I do not know about Kore.ai's scalability and whether it has met my needs as my project grows or changes.

I did not communicate with Kore.ai customer support, so I have no experience to share.

Before Kore.ai, I used Microsoft, and I did not switch to Kore.ai. However, we also tested Kore.ai because it offered many tools ready to test.

I encountered some minor problems in the setup for the proof of concept, but nothing significant.

My company does not have a business relationship with this vendor other than being a customer.

I have seen a return on investment from using Kore.ai, as it saved time for the implementation project.

I did not get access to pricing, setup cost, and licensing information because my organization handled that for me in my previous work.

Before choosing Kore.ai, we also tested Amazon AWS , but it was not very effective.

I do not want to add anything else about the features, including integrations or analytics, as there is nothing else that stands out.

My advice to others looking into using Kore.ai is that they can try it because it can be the solution. I would rate this product a 7 out of 10.

Our main use case for Kore.ai  is developing four chatbots which are used to serve customers relating to various tax products. We had different tax products in our organization, and our purpose was to provide customers with FAQ answers, obtain their bank details, retrieve their other product details, assist if they would like to order something, connect them to an agent, and answer any kind of query that they have regarding their products or anything tax-related.

One of the main chatbots I built for a co-system tax product allows users to come in and filter based on which product they are using. They can get information about their locators. If they have filed a tax and received some kind of rejection, they can get their information on that. We also had different kinds of forms which the user could fill and submit their information to the back-end team to act on it. Apart from that, they could also connect to a live agent to get any queries sorted which were not handled through the chat agent. We also had a case creation feature that creates a case at the back-end for someone in our team to look into.

Kore.ai  has positively impacted my organization by being a big money saver because before having our chatbot, the use case was that every time some of our customers needed help, they used to call a number. During tax season, the wait time on that number would go up to one or two hours, which cost our company a lot of money. Integrating these chatbots helped reduce that wait time significantly. Mostly, people can chat instead of calling, or they don't even need a person to solve their basic queries, which the chatbot can handle. It reduced the calls made to our agents by 50 to 60 percent. That was evident as our team, which started with three or four people for one bot, expanded to around five to six bots, and a lot of new people joined. We received a lot of extra funding to create agents for all the other products and teams we support.

Some of the best features in Kore.ai include that it is a low-code, no-code tool. I have seen a lot of low-code tools that don't give anyone any room to work with if they have a custom use case or something they don't support. Kore.ai is different in that way; even though it is a low-code platform, we can still write our code if we want and override whatever they are trying to do, in case it is ever needed. Apart from that, their Web SDK and BotKit support is very good, as they have all the functionalities, packages, and everything up to date and available for us to use. Also, they have a lot of in-built integrations for live agent connections. So if you want to connect with Salesforce , Genesys , Live Help Now, Telephony, or any of these kinds of back-end services, including ServiceNow , they just have an integration built-in, so you do not need to write it from scratch. But apart from that, they also give the ability to write it from scratch through a custom BotKit if you don't want to go with their features or if something is missing.

Regarding how Kore.ai can be improved, I think agentic AI is booming. Instead of us designing the whole conversation, which feels kind of boring and repetitive, they could move into a different kind of approach where we could write a prompt, and it should create a conversation based on that, in the way that Microsoft Copilot Studio  does. So they should move a little more towards the agentic AI approach and start working on and integrating different prompts so we do not need to design conversations. Nowadays, people are used to talking with ChatGPT, which facilitates a human way of conversation, just back and forth. People are not used to seeing a lot of menu buttons, icons, or lists to select things from. Kore.ai does a little bit of that using the agentic nodes and other features they have added, but they should move further in that direction.

The documentation for Kore.ai often poses a challenge because many times when we had issues, we couldn't find anything in the documentation. We either needed to figure it out ourselves through trial and error, or we raised a support ticket. Some of the engineers came and told us that we could do something to solve our problem. It took them two minutes to tell us, but when we asked if there was documentation for this, they mentioned there wasn't. They just knew it because it is some kind of internal knowledge. So if they improve their documentation, it will be beneficial for everyone, as we will not have to raise tickets for any issues.

I have been using Kore.ai for the last four and a half years.

Kore.ai is pretty scalable because when we started, we just had one bot. However, as we expanded to multiple bots with different kinds of BotKits and Web SDKs, we didn't face any issues with scalability. We also have many lower-end bots, so scalability has never been a problem.

Customer support is okay, I guess. We get what we want from them whenever there is an issue or something. But sometimes when we had issues, it was delayed a lot. We had to reach out to their support team, and while they tried to suggest solutions, they didn't always work. Eventually, after a few calls, they involved some engineers who could actually fix our issues. Sometimes it was bad, but I can say around 60 to 70 percent of the time, it was good. We got the support we needed.

I didn't have a previous solution because we have been on Kore.ai from the start.

I have seen a return on investment with Kore.ai, as it definitely has saved time. We previously only had a helpline where one agent could only attend to one customer at a time. But now, because it is chat, one agent can handle up to three or four chat sessions at the same time for different employees. That is certainly an improvement. Additionally, our chatbot solves a lot of users' problems on its own, meaning it doesn't even necessitate a live agent or person sometimes. This represents a direct time saving for agents to focus on more complex issues.

Before choosing Kore.ai, I know they evaluated other options, but that happened before I joined the team, so I was not part of those conversations. I know they considered Google's solution, which I forgot the name of, and also Yellow.ai and Google Dialogflow , but ultimately they went with Kore.ai.

One of the unique or unexpected things regarding our main use case for Kore.ai is that I have seen a lot of our teams just use the basic default template that they have. But because our company had a design system, we had to rewrite the whole Web SDK code and design it according to ourselves.

Apart from that, we also had to make a lot of changes in the back end to support a live agent because our live agent handles up to 20,000 to 30,000 users on a daily basis. We had to increase capacity and ensure everything was worthwhile so it did not fail while users are currently there, especially during the tax season.

My basic advice for others considering Kore.ai is that they are adding a lot of new features. They have integrated agentic nodes, switched to the XO 11 platform, and introduced GenAI features. It is beneficial to look into these developments and keep up with whatever is happening. Don't get stuck in old ways of making bots. They are introducing a lot of new features; some are good, some are not, but eventually, everyone should try those features and move towards the new agentic platform. I would rate this solution a 7 out of 10.