🧩 AI Files
Once you’ve prepared your datasets, the next step is to create an AI File.
An AI File is a .jsonl (JSON Lines) document that contains all your question–answer pairs in the correct format for fine-tuning OpenAI models.
These files are automatically formatted and stored both in Ainisa and on your OpenAI account.
📋 Overview
AI Files are used during fine-tuning to train your custom models.
Each file contains one or more datasets you’ve created manually or automatically from uploaded files.
You can create a new AI File from the AI Files section:
👉 https://ainisa.com/business/ai-files/create
🪜 Steps to Create an AI File
- Go to Create AI File
- Select Fine-tuning Type
- Choose the type of fine-tuning this file will be used for:
- Supervised — the model learns directly from good answers.
- Direct Preference Optimized (DPO) — the model learns from both good and bad examples.
- Choose the type of fine-tuning this file will be used for:
- Include Manually Added Questions
- Choose Yes if you want to include datasets you’ve created manually.
- Choose No if you only want datasets generated from your uploaded files.
- Select Files
- Choose the files (previously uploaded in the Training Files section) whose datasets will be added to this AI file.
- Click Save — Ainisa will create a
.jsonlfile containing your training data.

💾 What Happens Next
- Ainisa automatically converts the selected datasets into a
.jsonlfile. - The file is securely saved both on Ainisa servers and your linked OpenAI account.
- This
.jsonlfile will later be available in the Fine-tuning section to start your training job.
🧠 Notes
- Only OpenAI models currently support fine-tuning.
- A minimum of 10 datasets is required to create a valid AI file.
- The
.jsonlfile is generated automatically — no manual formatting needed. - You can create multiple AI files for different projects or fine-tuning types.
- If you include DPO datasets, make sure they contain both “good” and “bad” answers.
- Creating
.jsonlAI files is available starting from the Pro subscription plan.
See the Pricing page for details.
✅ Tip: For more details about fine-tuning methods, check out:
- OpenAI Fine-tuning Models
- Fine-tuning Best Practices (Video)
- Fine-tuning Best Practices (OpenAI)
- Supervised Fine-tuning
- DPO Fine-tuning