What is a Custom Model?
4 mins read - Created on Oct 24, 2025A Custom Model in Kimola is an AI model that you create and train with your own data. Unlike pretrained models that are built for general use, a custom model is designed to understand your specific categories, language, and context.
It allows you to teach Kimola’s AI exactly how you want text to be interpreted — making it ideal for brand-specific, industry-specific, or research-specific projects.
Sign in to your Kimola account and go to the Models section on the left panel. Select the Custom tab.
How a Custom Model Works
Custom models learn directly from the examples you provide. You upload your dataset (such as customer feedback, survey responses, or support tickets) and label each piece of text with the correct category. The AI then uses these labeled examples to understand the patterns behind your data.
If you’re analyzing employee engagement survey comments, you can label responses as work-life balance, communication, career development, or management support. After training, your model will automatically identify these themes in new survey results — even when employees express them in different words, such as “no room for growth” or “my manager always listens.”
Once training is complete, you can use your model to automatically classify new text data coming from reviews, social media posts, or any data source you connect to Kimola.
Ways to Create a Custom Model in Kimola
Kimola offers three different ways to create your own AI model, depending on your workflow and data type:
1. Create with a Prompt
You can generate a model simply by describing what it should do in natural language. Kimola will use your prompt to define the model’s logic, categories, and intended purpose automatically.
Create a model that classifies social media posts about your brands into categories such as environmental responsibility, ethical behavior, inclusivity, and community impact.
This method is ideal when you don’t have a dataset ready but want to quickly define a new classification logic. To learn more about creating models with prompts, see the Create a Custom Model with a Prompt article.
2. Create with a Training Dataset
If you already have labeled data, you can upload it as a Training Dataset.
Kimola will use your dataset to train a model that learns from your examples.
Upload a Excel file where each row includes a text (e.g., customer review) and its corresponding label (e.g., “Delivery”, “Support”, “Price”).
This option is best for researchers or teams who already have manually categorized data and want maximum accuracy. For a detailed guide on this process, see the Create a Custom Model with a Training Dataset article.
3. Create with a Report
You can also create a model directly from an existing Report in your Kimola dashboard. If your report already contains labeled or categorized records, Kimola can use that data to train a new Custom Model automatically.
If you’ve categorized consumer comments in a previous study, you can turn that labeling work into a reusable model to automate future analyses.
This method is ideal when you’ve already completed a classification task within a report and want to use that labeled data to train a model without starting from scratch. To learn more about this feature, see the Create a Custom Model with a Report article.
Why Create a Custom Model?
Creating a custom model gives you full control over how your data is interpreted.
It’s especially useful when your categories or terminology don’t match those used in Kimola’s pretrained models.
- It understands your brand’s unique tone and vocabulary.
- It performs better for niche industries or specific research fields.
- It helps automate repetitive manual labeling tasks.
- It improves accuracy over time as you retrain it with new data.
When to Use a Custom Model
- You have data from a domain that isn’t covered by Kimola’s pretrained models.
- You need to classify text using your own set of categories.
- You’re conducting specialized research or working with internal business data.
If your data already fits one of Kimola’s AI Models — like E-Commerce Feedback Classifier, Sentiment Classifier, or News Classifier — you can start analyzing right away without training a new one. Otherwise, building a Custom Model will give you more accurate, domain-specific insights.
