FAQ

General questions

  • What is a chat platform?
    • It is designed to facilitate instant and interactive conversations between businesses and their website visitors or app users. After the chatbot redirects users to the agents, they start to communicate between each other, the agent from chat platform, the user from chatbot.
  • What is a chatbot platform?
    • A chatbot platform is a software or service that enables the creation, deployment, and management of chatbots. 
  • What is an external request?
    • An external request refers to a request made by a chatbot to an external system or service to fetch or process data.
  • What is required to integrate with the platform?
    • Rest API documentation. If the platform is popular, then most likely we already have a configured connector.
  • What is needed to start?
    • It is necessary to fill out a brief with the client to obtain key information, in addition, it is necessary to understand what type of chat bot and for what use case the client wants to receive, as well as key project success metrics.
  • What is an orchestrator bot?
    • An orchestrator bot is a chat bot that manages different bots. As an example, if a customer has a sales bot, a FAQ bot, and a customer service bot, the Bot Orchestrator will allow the user to seamlessly navigate through these bots based on the request.
  • Do I need to create separate chat bots for each messenger?
    • No, the owner has to create an one chatbot in GMS chatbot platform in Bot Builder and connect it for necessary messenger.
  • What is a no/low code platform?
    • Our Chatbot platform allows customers to create chatbots without having to code. Bots are built from pre-defined blocks and the client only needs to add the necessary functions. Also, to create a bot with NLU technology, the client can register intents and training sample in Excel and upload the file directly.

Channels questions

  • Can the interlocutor conduct a dialogue with the bot, seamlessly switching between different channels (Telegram, Viber, WhatsApp, etc.) and devices? For example, a client communicated with a bot via Viber from a phone, then logged into another messenger from a laptop, will the bot be able to recognize the user in this case and continue the previously started conversation in a new channel?
    • For example, a client communicated with a bot via Viber from a phone, then logged into another messenger from a laptop, will the bot be able to recognize the user in this case and continue the previously started conversation in a new channel?
  • Will the history of the user’s actions on the bot be available when contacting the operator?
    • Yes, the history of the user’s actions will be able to be available. If it will be in the Technical Requirements from customer.
  • Does the ability to independently close the chat by the bot user?
    • Yes, the bot user can close the chat after solving the problem or if it is not needed anymore. Also it must be added to the dialog design.
  • Will the test bot be deployed as the customer has it now (improvements – then in prod)? We plan to develop it – therefore, all improvements must be implemented on the test bot.
    • Test bot can be deployed on the Telegram or Viber to internal testing. Also the dialog designer can test a changing bot on the Chatbot platform in the debug mode.

Data & Security questions

  • What level of security do we provide to the client when we connect to his CRM? How can we influence the leakage of confidential data? Will hackers be able to get sensitive data either by disguising themselves as a specific user or by hacking a chatbot to provide all sensitive data from CRM?
    • Security is provided by the CRM provider. If an external IT system (customer’s CRM system) involves integration over an insecure http protocol, then we do not guarantee anything. When using secure https protocols, we provide everything that is said in the response.
    • In public channels, when working with confidential information, you can idle a script that will not give the user access to the content if he (the user) has not passed additional authorization. If an attacker has taken possession of the user’s login\password\device and communicates with the chatbot on behalf of the user, we will not know this in any way.
  • Who has access to customer CRM data? Will they be stored on our servers?
    • The CRM itself resides on its own servers and stores data there. The data of the clients that the bot uses will be stored on the server where the chatbot platform is located. They can be accessed by developers and dialog designers if needed.
  • When a bot initiates a conversation, does the client need to provide a proactive notification to end users, or are there any legal issues that could cause?
    • Only in WhatsApp it is possible to initiate a conversation before the chat is activated, for other channels the end user must activate the chat to allow the bot to initiate the conversation. New clients: if a client (a future user of a chatbot) signs a document (for example, consent to the processing of personal data), it is necessary to add a clause on consent to receive mailing lists (indicate mailing channels) Current base: consent must be obtained in the first WhatsApp message sent, for example: “Do you agree to receive informational messages from us?” / By sending a message, you agree to the processing of personal data in accordance with our policy.”‘
  • Where will the chatbots be placed, in cloud services or on servers?
    • Depending on the requirements of the customer and the budget of the project, we can host a chatbot on a SaaS server or deploy a platform in the Customer’s environment and create bots on it.
  • Can a chatbot be hacked as a result of fraud when an irregular number of requests are sent at the same time?
    • No, because there is protection against such actions. There are monitoring and certain restrictions. If abnormal behaviour is observed, we will stop the chatbot and notify the customer.
  • User Authentication via Chatbot: do chatbots support user authentication to ensure secure and personalized interactions?
    • We could provide multiple types of user Auth and design auth scenario by channel / scenario / CRM segment / security requirements. In our platform we just need to create a chat / visitor ID and match it with your CRM and then route the user on the corresponding chatbot scenario. Flexibility in orchestration is one of key features in our platform.
  • How secure is the user’s personal data?
    • In public channels, when working with confidential information, you can idle a script that will not give the user access to the content if he (the user) has not passed additional authorization. If an attacker has taken possession of the user’s login\password\device and communicates with the chatbot on behalf of the user, we will not know this in any way.

Dialog Design & Integrations questions

  • How long does it take to form a new language?
    • The minimum is 2-3 weeks, the maximum is 1 month. Available languages: Ukrainian, English, Kazakh, Spanish, Portuguese, Vietnam, Urdu, Indonesian, Malaysia, Filipino.
  • Can the bot insert screenshots if necessary? Do you mean taking pictures and processing them?
    • Yes, the chatbot can work with media files. But there are restrictions on working with media from certain end channels.
  • Can we set a certain tone of communication that suits us best depending on our niche?
    • Yes, the chatbot can be personalized.
  • Chatbot and Direct Chat: do you have examples of how users can seamlessly transition between using a chatbot and connecting with a live customer support representative. Do you configure the chatbot integration process to meet customer-specific business needs?
    • Depends on a live chat platform and use cases. We are flexible in fine tuning for live agent handover, but some specific cases could be limited by a live chat platform.
  • Integration Capabilities: do you provide integration capabilities with existing customer support interfaces, web and mobile platforms? Are you able to customize the UX/UI according to customers’ brand guidelines.
    • We have multiple integration frameworks: for channels we have sync & assync API, for IT systems we have API builder. Customization of UX / UI based on the branding guidelines is a standart practice, we could provide you with a guidance on how to implement it.
  • What does it mean to build dialog rule-based scripts and configure navigation through the button menu?
    • Building dialog rule-based scripts and configuring navigation through a button menu for chatbots refers to a method of designing conversational flows and user interactions in a chatbot system. It involves creating a set of predefined rules or decision trees that dictate how the chatbot responds to user inputs and guides the conversation. Overall, building dialog rule-based scripts and configuring navigation through a button menu are techniques employed to create a more controlled and user-friendly chatbot experience, ensuring that users receive accurate and relevant responses while simplifying the interaction process.
  • What is the deployment scenario for On-Premise solution?
    • Deployment process: Infrastructure preparation, Transferring a distribution kit with docker images and platform components, deployment of the platform according to the instructions, Deployment of infrastructure services, Setting up environments according to the agreed CI/C process, Testing and Debugging.
  • What platforms can the chatbot integrate with?
    • Messenger: Telegram, WhatsApp (dialog360), Microsoft Teams, Viber. Omnichannel: Omnichannel, Edna Chat Center, LiveTex, Chat2Desk.
  • How to set up an external request?
    • The External Request slot does not contain settings for the HTTP request it sends. It contains a link to the External Request from the Company Resources, and the External Request in its turn contains all the necessary settings for sending the request and processing the response.  Name is the name of the slot. The maximum length of the field value is 1000 characters. Once the maximum length is reached, no more characters can be entered into the field; Description is the description of the slot. Maximum length of the field is 1000 characters. On reaching the maximum value no more characters can be entered into the field; Request method is the HTTP method for this request: GET POST PUT PATCH DELETE HEAD Endpoint is the address of the webhook to which this request will be sent. The maximum length of the field value is 1000 characters. When the maximum value is reached, no more characters can be entered in the field. 

NLU & Supported languages questions

  • NLU Quality: What is NLU quality? Are there any processes you have in place for continuous NLU improvement?
    • We will measure F1 Score and could customize NLU pipeline to achieve the best quality adjusting pre-processing or vectorization components. At the same time we are working on NER (named-entity recognition) to increase NLU recognition. Normally, if the architecture of the chatbot is designed professionally and training dataset is ok, we could achieve 90%+ accuracy.
  • What is bot personalization (small-talk intents)? For this we need to agree with you on intents and responses to them?
    • Bot personalization refers to the customization and adaptation of a chatbot’s responses to create a more personalized and engaging conversation with users. Small-talk intents are designed to provide a more human-like interaction by incorporating social elements, humour, empathy, or general chit-chat into the conversation. These intents cover a wide range of topics that are typically unrelated to the primary purpose or functionality of the chatbot. They can include greetings, compliments, jokes, weather discussions, sports updates, and other casual topics that help establish rapport and make the conversation more engaging.
  • What is NLU?
    • NLU is a Natural Language Understanding. On the platform the DD can implement speech recognition of the Interlocutor by an Agent for processing messages in natural language in order to determine the expressed intention (Intent) of the interlocutor and, depending on the specific intent, make the transition to the appropriate branch of the Script.
  • What is Waiting for Reaction?
    • Wait For Reaction is a slot in Bot Builder that puts the Agent in the waiting mode for the Interlocutor’s message and saves the received message to the System context variable client_message. 
  • What is Jump?
    • Jump is a slot in Bot Builder for moving to another Slot.
  • What is an intent? How to use it?
    • Intent — the intention of the Bot User, the theme of the Bot User’s utterance which the Bot is supposed to recognize. Each NLU slot in the Script is trained on the training sets of those Intents which are included in its Intent sub-slots. The rest of the Company’s Intents are not involved in recognizing utterances in this NLU slot. However, they can be used in another NLU slot. 
  • What is a Vocabulary? How to use it?
    • When expressing intention in natural language, the Interlocutor can use different Entities in the phrase. For example, in the phrase “How to get to NYC by train” there is an Entity “city” (NYC) and “transport” (train). The Entities can be expressed by different synonyms. For example, “NYC” can be called “New York”, or “NY”, and “train” – “train” or “railway”. Some scenarios require you to select Entity values and write them to Chat Context Variables for later use in the Script.
  • Where to start creating a bot?
    • The user can start creating a bot in Agent Design or Bot Builder on the GMS Chatbot platform.
  • Can the chatbot handle bilingual communication?
    • Yes, the chatbot and chatbot platform can handle bilingual communication, if it is needed. Also the chatbot , Ukrainian with dialect or English slang etc.
  • What is a training sample? How to set it up?
    • The training sample is a list of query differences used to train NLU. The user can configure it either directly on the platform in the Intents tab, or by writing it in an Excel file.

Analytics questions

  • What does basic and customized analytics look like, who generates them, and how often?
  • Can we count the funnel for marketing companies?
    • Yes, we generate marketing funnels and other customized analytical reports on request, it is important to collect requirements before the start of the project, as it may be necessary to adapt the dialogue design to the requirements for analytics.
  • Automated calculation of NPS
    • Unfortunately, we don’t have an automated calculation of NPS, but we can do it on our end manually and send the customer a report every week/month/quarter/year based on the agreement.
  • Is it possible to send questionaries aftrer closing the chat with users?
    • Yes, the Chatbot and Chatbot Platform can send questionaries and save it to send into analytics.
  • Chatbot Analytics: What about tracking tools to measure the effectiveness of a chatbot and improve its performance over time?
    • We are tracking everything – all logs, and based on this logs we could analyse the behaviour of user in BI systems or build any custom report. Our Product team is working on real-time analytics and visualization, we plan to release some features in 2023.