Customer service agents spend most of their time solving simple but repetitive customer queries. Our product is a software that provides automatic, intelligent response recommendations for customer support agents. We can integrate our recommendations with your existing CRM system. The product is more intelligent and effective than currently used automation alternatives, such as copy & paste templates or chatbots.
If you are facing a high volume of repetitive customer queries, the highly probable answer is yes! If you face challenges in optimising your support function, please reach out. We would love to help you!
Yes, the AI learns how the customer queries are addressed by your agents by going through your past customer interactions, regardless which industry you are in.
Our software supports European languages (we have successfully worked with English, Spanish, German, Italian, French and Finnish). If you have are interested in using a particular language, please send an email to firstname.lastname@example.org with details.
We support all text-based channels, such as live chat, social media and emails.
We charge a bespoke fee per used suggestion or a fixed-term license. Please contact us for a quote.
We can show you a demo of our product trained on a sample dataset. Contact us at email@example.com if you are interested in a demo. If this seems like a good potential fit, we can then set up a demo customised to the data and customer service questions in your company.
We are different in two main ways:
With our solution we let the AI present several best recommended answers to the agent. The agent can then choose which response is best and send it. This drastically reduces the time it takes for agents to send a reply.
A more common alternative would be to let the AI automatically send the response (chatbot). If the response chosen by the AI is not good the customer will have a poor user experience. By presenting recommendations to the agent, the agent is still responsible for ensuring a great customer service experience.
Yes. When training the AI it will learn from your experienced agents. This results in the AI suggesting replies that leverage this experience. The new agent will in effect be passing on the responses of an experienced agent, raising the quality of the customer experience, while simultaneously teaching the new agent the best answers.
We use a cutting-edge AI which is built on end-to-end machine learning technology. We let the AI go through all your historic conversations. This way, it understands what a good answer to any given question are. The patterns learned from the initial training allow the software to generate relevant response recommendations in future.
End-to-end machine learning is what we use at True AI. The idea is to use machine learning to understand the entire conversation. It models both the parts of the conversation, and how those parts are put together by using machine learning. The AI is taught not just what to respond to a particular question (as in chatbots), but also how the dialogue preceding the final question influences what answer it should give.
Partial machine learning is what is usually used to create modern “chatbots”. The chatbots use IF-THEN type rules ("IF the customer says X, THEN respond with Y") and uses machine learning (maybe even neural networks) to make the “IF” part a little more flexible.
Partial machine learning uses machine learning to understand parts of the conversation, and relies on explicitly programmed rules to understand how those parts are connected. It works well for extremely simple conversation (where there are only a few things that can be said or requested) but fails for more complex conversations.
Partial machine learning only relies on machine learning to understand parts of the dialogue, with the remainder requiring multiple IF-THEN rules to function. If you have anything but an extremely simple use case in mind this will become very complex and expensive to engineer and to later maintain and update.
End-to-end machine learning on the other hand is much cheaper, only relying on past historic conversations to make the AI understand the flow of the conversation. No rules for deciding how the conversation should flow need to be implemented.
We follow strict security policies and best data safety practices, including but not limited to Microsoft Azure threat blocking, detection and logging services, Azure Web application firewall for detection and prevention of malicious requests, Azure DDos protection, HTTPS encryption, storage encryption and anonymisation, OAuth2 authentication, Azure KeyVault storage, employee security and confidentiality training, security clearances and NDAs, professional indemnity as well as public liability insurance.
We have currently developed, or are planning to develop integrations for Salesforce, Twitter, LivePerson, ConverSocial, Bold360, Hootsuite, Intercom, Microsoft Dynamics, Zendesk, Sprinklr, TweetDeck, Buffer, and FreshDesk. We are happy to build a custom integration with most systems, which is a process that takes ~2 weeks.
We have a product API. Please contact firstname.lastname@example.org for a preview.
Depending on the dataset we will need between 5,000 - 100,000 conversations to create an effective solution.
We charge a bespoke fee depending on how complex the integration procedure is.
After we have received the data, we will need between 5 - 15 days to prepare our AI solution by training the algorithm.
We will provide you a secure upload link offer technical support throughout the data sharing process.
JSON with as much detail as possible is preferred. If that is not convenient, we can handle almost all other data formats (CSV, XML, etc.).
True AI Ltd. is registered in England & Wales, company no. 09864737. Registered address: 4 - 5 Bonhill St, London EC2A 4BX, United Kingdom.