Company Assistant can advise your client in private messages on common key objections. It is an Artificial Intelligence that closes objections and even leads the client to payment without your involvement. Perfect for Instagram and Telegram businesses.
How does AI-template work?
This template is designed to respond to any Instagram Direct or Telegram bot message, without a specific trigger.
Each user message is overwritten in the variable "question". The Neural Network will generate a response to the user message stored in this variable:
The intermediate message block with the text "..." is needed to keep the client's attention and to show the user that the answer will come soon, because the Neural Network generates the answer not instantly, but within a few seconds, depending on the load.
The most important block in this chain is the Neural Network.
With the help of an assigned role, the request to the Neural Network will be more precise, and the generated response will be more specific. We recommend specifying a role for any type of request.
Example 1: Take the role of an AI assistant who helps answer math-only questions.
Example 2: Take on the role of a professional copywriter who creates content for various platforms such as Instagram, Twitter or YouTube.
Example 3: Take on the role of an AI-assistant on @dima.torgov's Instagram account. The main goal is to respond to Direct messages on my behalf (as if you were Dima). Respond in the same language and tone of voice as in the question.
How do I train a Neural Network to suit myself?
The knowledge base is the request to the Neural Network, i.e. what exactly the Neural Network should do, what actions to perform. For example: solve equations, generate numbers, phrases, texts for publications, answer questions, the query can be anything.
Write a request to the Neural Network in as much detail as possible, specifying all the particulars.
It is recommended to fill the knowledge base in English, it will increase the quality of answers and reduce the time of answer generation.
Neural Network's answer is saved in a variable, so that further in the message you can display the values of this variable and the user will get the answer to his question:
Additional settings for the Neural Network’s action.
1. Without message history:
If you do not care about the message history between the client and Neural Network, we recommend enabling this option. This will noticeably reduce token consumption.
2. Optimize all requests:
ChatPlace's smart algorithms know how to optimize requests to the Neural Network to save tokens. We recommend enabling this option only in the case of a dialog between the client and Neural Network when using your Knowledge Base.
After the user writes any message - this automation starts from the beginning, the user's message is saved to a variable, the Neural Network reacts to the message saved in this variable and generates its own reply which will be sent out in a second variable.
How to start the Assistant?
1. Select the "Any Message" start trigger in the top right corner.
For Telegram:
For Instagram:
2. Go back to the "Automations" section and run this automation:
Trigger on a specific trigger word.
This template is designed to have the assistant communicate with the client without your input, so the trigger is customized to “Аny message”.
In order for the AI-assistant not to run in parallel with the live operator - you will need to change the beginning of the chain a bit and set a specific word or phrase as a trigger.
To launch you will need to:
1. Create a welcome message so that the client realizes that they are communicating with the AI-assistant and make this message the start of the chain by clicking on the arrow above the block
2. After that, create a new Action block
3. Select the action - “Collect client data”
4. In the action settings, activate “Text” and “Save answer to variable”
5. Select the variable from the dropdown menu to which the user's question will be saved, in our case it is the “Question” variable
6. Enable the option: “If client responded after”
7. Set a small amount of time, 5-10 minutes.
8. Enable the “Exit and send error message” setting
9. Write a text message in case the client responds after the wait for the client's message is over
This is to prevent the assistant from waiting forever for the client's message and interfere with the communication between the client and the live operator.
10. Now just connect the first message with the resulting Action block.
11. And for the assistant to be able to answer any new questions from the client - connect the block with the assistant's answer (with the {{ answer }} variable) to the "Collect client data" action:
Done! Now set a trigger for this automation, e.g. "Assistant"
The automation is ready to run!
Additionally, in the Growth Tools section, you can set up a Welcome button to start this automation so that new users opening your Direct for the first time can easily launch the assistant without even knowing the trigger word.