Step 1: Call the AI agent in the Telegram chat

1. Log in to ASCN.AI and create a new WorkFlow.

2. Add Trigger → Telegram → Trigger Telegram Polling

3. Configure data:

BotToken: Here you need to insert your bot's token, you can get it in @BotFather

4. Add Tools → Storage → Storage Put

5. Configure data:

Key: summarychatbot (The key can be anything, the main thing is to use the same one everywhere for correct data processing.)

Value: here we transfer the value we want to save

= {{ {{ JSON.stringify({
  text: "User Message: " + ($json.update.message.text ?? "—"),
  user_id: $json.update.message.from.id ?? null,
  username: $json.update.message.from.username ? "@" + $json.update.message.from.username : ($json.update.message.from.first_name ?? "Noname user"),
  created_at: new Date().toISOString()
}) }}

6. Add another node in parallel (from the Trigger Telegram Polling block) Logic → JavaScript → JavaScript Run

7. Configure data:

Code: Here we write the code that will be executed when the node is activated

=// Get text from Telegram Polling
const update = $json.update ?? {};
const message = update.message ?? {};
const text = (message.text ?? '').trim();

// Checking for the presence of the phrase “@summascnbot”
const containsBotTag = /@summascnbot/i.test(text);

if (containsBotTag) {
  return { text, status: 'allowed' };
}

throw new Error('Сообщение не содержит упоминания @summascnbot');      

Expose Context: True, When activated, content (variables from the workflow) can be added to the code

8. Add (after the JavaScript Run node) Tools → Storage → Storage Get

9. Configure data:

Key: summarychatbot (The key must match the one entered in the second node)

Step 2: Configuring the AI agent for chat analysis

1. Add AI Agents → Agents Prompt

2. Configure data:

UserPrompt: This is your prompt, which you send to the AI model, and you can also send the necessary data here

={{$node['Storage.Get_1'].json}}

System: Here we specify the system data for the AI model (its role, limitations, formats, etc.)

You are an AI analyst at a company. Your task is to summarize the daily communication in the work chat. Analyze all messages and compile a structured summary of the day. Be sure to highlight: 1. Main topics and discussions. 2. Key decisions or tasks that have been agreed upon. 3. Mention the names or usernames of participants if it helps to understand who raised the topic or made the decision. 4. Describe the overall dynamics and mood of the team. Write concisely, without introductory phrases or evaluations. Do not mention technical messages, greetings, stickers, or one-word responses such as “ok” or “got it.” The style should be businesslike, neutral, and readable as a finished report. DO NOT USE MarkDown FORMAT!

Provider: Openrouter (for Deepseek R1) select API Token provider for AI model

Model: openrouter/deepseek/deepseek-r1 (select the model itself)

API Token: Insert your AI token here. You can get it on the official website

Temperature: 1 (from 0 to 1; the higher the value, the more creative and imaginative the AI model will be)

Step 3: Displaying information in the chat

1. Add Tools → Telegram → Telegram sendMessage

2. Configure data:

BotToken: Your token bot, which was in the first node

ChatID: Your Telegram chatID (or chatID group/channel) where summary reports about the chat will be sent. We recommend choosing either a group (where you will initially add the bot) or your own idUser

Text: Here we write what exactly will be sent to the user. In our case, the result of the AI agent's analysis

={{$json.text}}

3. Add Tools → Storage → Storage Remove

4. Configure data:

Key: summarychatbot (the key is the same as in Step 1)