🎯 1. Strategy and Business Context
The Problem
Many companies need their virtual advisors to answer inquiries using information that constantly changes (vehicle inventories, product catalogs, prices, or property availability). Keeping this content up to date manually consumes valuable time, generates outdated responses, and creates operational overload for human agents.
The Solution
The Smarton Dynamic Tables feature automatically connects your databases with your chat channels. The system absorbs, cleans, and organizes information periodically, allowing the bot to respond in 100% natural, fluent language with data verified in real time.
☝️ When to Use It? (Ideal Customer Profile)
It is recommended to choose this solution if your operation meets the following characteristics:
Inventory is live and dynamic: Information changes daily, making it unacceptable to give an incorrect price or sell something out of stock.
High product volume: You have live catalogs of more than 100 products in stock.
Data centralized externally: Your critical information is already organized in your own system, a CRM, an ERP, or a spreadsheet.
Complex commercial searches: Your users typically ask questions based on budgets, brands, models, or years, and you need the bot to understand ranges and proximity (for example, searching for a car "around $15,000").
🤖 How to Connect and Configure Your Smarton Step by Step?
Configuring your Smarton requires connecting your "source of truth" and teaching the Artificial Intelligence how to read your data. The system extracts, cleans, and stores information autonomously without human intervention.
General access route: To get started, go to Settings > Resource Manager > Dynamic Tables in your side panel and press the "New Table" button.
In the creation wizard, the first step is Selecting the data source. Depending on your company's infrastructure, you can choose between two options:
🟢 Option A: Google Sheets Integration
Designed for sales teams that manage their operations in spreadsheets and do not have a technical department.
Secure login: Log in with the Google account where the Google Sheets file you want to query to bring your table is hosted.
File and Sheet Selection: The system will list your available documents. Select the desired file and then choose the exact sheet (tab) of that file that you will use to feed this table.
⚠️ Mandatory Format Rules: For the bot to understand the information, your table must start mandatorily in the first cell (A1), must not have merged cells, and has a maximum limit of 1,000 rows by 50 columns.
2. Define name, description, and frequency: * Name and Description: Assign a name to your connection. It is vital that you add as much detail as possible in the description, as this gives the AI the necessary context about what your table is exactly about.
Update frequency: Choose the "heartbeat" of your inventory (how often the system will search for new changes). You can select: Every 6 hours, 12 hours, daily, weekly, or every 15 days.
3. Choice of Fields (Columns): Select the columns from your table that you want to import to the platform and assign them a clear description.
Important: You must be very specific in the description of each column, as this information will be analyzed by the AI to extract the most accurate answer possible for your customer.
4. Create and Connect: Once everything is configured, press the "Create" button to start using your database connected directly to your file and ready to be linked to your Smarton.
🔵 Option B: API Integration (Your Own System / CRM / ERP)
Designed for technical teams that already have their own database, CRM, or structured system and want to connect and index it directly.
Source Selection: In the table creation wizard, select the API Tables option.
Define Basic Data and Frequency: * Name and Description: Assign a clear identifier to your connection. Write a detailed description; remember that this text gives the AI the initial context about what type of information it will extract from your system.
Update period: Define the automatic time interval in which Atom will consult your API to keep the inventory up to date (Every 6 hours, 12 hours, daily, weekly, or every 15 days).
API Call Configuration (Request):
HTTP Method: Select the type of call corresponding to your service, either GET or POST.
Endpoint: Enter the exact URL of the server or external system where your structured inventory information resides.
Authentication, Headers, and Body:
Authentication Method: Configure the type of access according to your company's protocols (simple authentication with username/password or authentication via API Key/Bearer Token).
Headers: Add the mandatory headers required by your service (for example,
Content-Type: application/jsonor custom security parameters).Body: If you use a
POSTmethod or require the system to send a specific JSON to filter the catalog, structure the body of the request in this space.
Test Request and Response Codes:
Execute the test button integrated into the wizard to perform a call in real time.
Validation: Carefully review the response code (Status Code) returned by your server. An OK will confirm that Atom has connected successfully and has received the JSON of your inventory to proceed with mapping fields with the AI.
📄 Loading Additional Documents (Support Policies and FAQ)
Catalogs or databases solve the hard data of your products (prices, stock), but customers also ask mixed service questions ("Do you have financing?", "Where are you located?", "What is the warranty?"). To solve this in the same channel without leaving the sales flow, you can load text files (PDF or Word) that complement the operation of your dynamic table:
Immediate Processing: Unlike the table via API or Sheets that syncs under the frequency you programmed, these additional documents are processed in the bot's memory in an instant manner. The bot learns the information in the exact second you upload or replace the file.
Unified Responses: When a user makes a compound query, the Smarton consults both sources in milliseconds and delivers you a single fluent response.
Mixed example: "Hello! Yes, we currently have 2 Toyota Corolla 2022 available starting at $17,800 USD [Data extracted from your API]. Regarding financing, we offer plans from 12 to 60 months with a preferential rate of 8.5% annually [Data extracted from the support PDF]."
⚠️ Priority Rule: The structured data from your Dynamic Tables (API / Google Sheets) always has absolute priority as it represents your stock and prices in real time. Uploaded documents act solely and exclusively as supporting commercial backup.
🛠️ Core Functionalities and Data Intelligence
The Smarton includes advanced processing tools to ensure successful searches and eliminate errors:
📊 Automatic Spelling Correction
During each update, the system analyzes your columns. The bot is trained to understand what your customers are searching for regardless of typos, abbreviations, or formats they use when typing.
Actual data in your inventory | How the customer searches for it | Search result |
"CX-5" | "cx5" or "cx 5" | Found (Ignores hyphens and spaces) |
"TOYOTA" | "toyota" | Found (Ignores uppercase and lowercase) |
"Córdoba" | "cordoba" | Found (Ignores missing accents or diacritics) |
"White" | "blanco" | Found (Understands both languages) |
🛡️ Strict Rule: "Zero Invention"
Unlike other virtual assistants that invent data when unsure about something (hallucinations), this Smarton has a mandatory technical rule. The bot is strictly forbidden from assuming prices, inventing features, or taking for granted that a product is available. If the product does not exist in your database, the bot will openly say there is no stock before giving false data.
🧮 Commercial Intelligence (Flexible Ranges)
When a customer searches by budget, model year, or mileage, the system avoids requiring an exact match to prevent leaving the user without options. Instead, it applies flexible margins:
Prices / Investment: Range of ±10% (e.g. if the user searches for a car of "$20,000", the bot will offer options between $18,000 and $22,000).
Model Year: Range of ±2 years (e.g. if they ask for a "2015" model, it will evaluate options between 2013 and 2017).
Mileage: Upper limit of up to +15% (e.g. if searching for a car with "50k km", it will consider alternatives up to 57,500 km).
💡 AI Exceptions:
Best options first: The bot will automatically sort results to display the options that best match the ideal budget first.
Exact Search: If the customer explicitly writes the word "exact" or "exactly", the bot will cancel flexibility and search for that exact number.
⚙️ Instruction Configuration: Search vs. Presentation
The Smarton internally works with two independent "brains": one that searches the data and another that drafts the final message.
🛠️ Advanced Configuration (Search Brain)
What does it do? It teaches the system how to interpret the user and how to search within your columns.
What to configure here? Clarifications about what each column means or rules on how to filter your inventory. It does not define the bot's personality or speak to the customer.
Golden Rule: This configuration belongs to the table, not the bot. If you use the same table in three different Smartons and modify the Advanced Configuration in one, it will affect the other two.
🎨 Personalization Prompt (Writer Brain)
What does it do? It shapes the bot's personality. It receives the data that the search brain has already found and drafts it in a friendly manner.
What to configure here? The tone of voice of your brand (formal, fun), greetings, use of emojis, and how to display information (e.g. "Always present prices highlighted in bold"). It does not search for information.
IN SUMMARY: If the bot doesn't find data, adjust the Advanced Configuration. If you need it to speak more beautifully, adjust the Personalization Prompt.
🆚 Generic Smarton vs. Dynamic Tables
Feature | Generic Smarton | Dynamic Tables Smarton |
Source of information | Long texts, manuals, or informational PDFs. | Spreadsheets (Google Sheets) or own databases (API). |
Search Method | Reads texts searching for similarities, like a human reading a manual. | Searches in exact cells and columns, like a supermarket system. |
Risk of Inventing Data | Low, but sometimes tries to summarize or deduce if the text is unclear. | Absolute zero. If the data is not in the table, it doesn't invent it. |
Price Intelligence | It struggles to understand if a price is "close" to another. | High. Understands budgets, tolerances, and proximity of numbers. |
Maintenance | You manually upload a new file every time the information changes. | Automatic. It updates itself every few hours as you program it. |
✅ Choose Dynamic Tables if: Your inventory changes constantly, you have product volume, and you need absolute guarantees that the prices provided are 100% truthful.
📚 Choose Generic Smarton if: Your information is institutional and fixed (company policies, regulations) and you want the bot to explain texts instead of selling products.
🚦 Best Practices for Administrators
To ensure optimal behavior and prevent the bot from getting confused:
Keep instructions separate: Never tell the bot how to be "friendly" in the Advanced Configuration section, and never explain to it how to "search in the table" in the Personalization section.
Data Hygiene (One language only): Don't mix languages within cells of the same table. This severely confuses the AI when correcting words.
Functional Synonyms: If in your country they call vans "busito", make sure your table contemplates those synonyms so the bot can connect them easily.
Information Dosage: Avoid giant messages. Ask the personalization prompt to show a maximum of 3 results per message, inviting the customer to ask if they want to see more.
Handling Gaps: Ask the bot in the personalization prompt to elegantly ignore fields that are empty or have dashes in your table (so it doesn't say things like "Color: None").
Let Atom work for you! 🚀
The Dynamic Tables Smarton with support documents is the definitive solution to unify the precision of your inventories with the warmth of excellent customer service. Set up your synchronizations and watch how your business serves automatically 24 hours a day with transparent, commercial, and 100% reliable information. ✅