Create Your First Project in Kalemat AI CAT Tool

Starting your first translation project in Kalemat AI is the first step toward faster, more consistent localization work. Whether you are translating a single document or managing a multi-language campaign for a global client, this guide walks you through creating and configuring your project from start to finish.

Before You Begin

Ensure you have the following ready:

• An active Kalemat AI account with available AI words remaining in your monthly allowance
• Your source files (supported formats include .docx, .xliff, .json, scanned PDFs, and images)
• Source and target language specifications
• (Optional) A list of brand-specific terms or "do-not-translate" words

Tip: Check your AI word balance in the top-right corner of your dashboard. If you plan to use high-end AI models with higher word multipliers, ensure you have sufficient tokens for your document size.

Step 1: Start a New Project

1. Log in to your Kalemat AI dashboard
2. Click the "New Project" button in the top navigation bar or the center of your project list
3. Select "Translation Project" from the dropdown menu

Step 2: Enter Project Details

Fill in the basic project information:

• Project Name: Use a descriptive name (e.g., "AcmeCorp_ProductManual_March2024")
• Client: Select an existing client from your list or create a new one
• Source Language: The original language of your content
• Target Language(s): Select one or multiple languages for translation

Tip: If you select multiple target languages, Kalemat will create separate tasks for each language pair, allowing you to assign different translators or AI models to specific markets.

Step 3: Upload Source Files

Drag and drop your files into the upload area or click to browse your computer.

• For editable documents (.docx, .xliff, .json): Upload directly
• For scanned documents or images: Toggle the OCR Processing switch to extract translatable text while preserving layout

Tip: OCR processing consumes additional AI words but saves hours of manual retyping. It is ideal for old PDFs, signed contracts, or image-based content.

Step 4: Choose Your AI Model and Settings

Select the AI "brain" best suited for your content:

1. Click "AI Model Selection"
2. Choose from available models (Gemini, GPT-4o, or Kalemat Custom Model)
3. Review the Word Multiplier indicator next to each model

The Word Multiplier shows how many AI words each translated word will consume. Fast models may use 1x words, while advanced creative models may use 2x or 3x words for greater contextual accuracy.

Tip: Start with a lighter model for first drafts of simple content. You can always switch to a premium model mid-project if the tone requires more nuance.

Step 5: Configure Workflow (Business Plan)

If you are on the Business plan, set up your project workflow:

1. Click "Workflow Builder"
2. Select or create a workflow:
   • Simple: AI Draft → Final
   • Standard: AI Draft → Human Edit → Final
   • Advanced: AI Draft → Human Edit → Senior Review → Final QA

The system will automatically move tasks to the next team member when each stage is marked complete.

Step 6: Set Up Term Base Integration

Ensure consistency across your project:

1. Toggle "Enable Term Base" if you have existing glossaries
2. Select relevant term bases for this client or industry
3. Add new terms specific to this project (brand names, technical jargon, or "do-not-translate" items)

Tip: Terms added to your Term Base here will automatically sync to all team members if you have Global Translation Memory enabled on your plan.

Step 7: Configure Translation Memory

Kalemat AI automatically manages Translation Memory (TM) assignment to ensure your translations are preserved and reused across projects.

Automatic TM Creation and Assignment:

• Each translation task automatically creates a dedicated Translation Memory or assigns to an existing one based on your client and language pair
• This TM is set as the default memory for your project, ensuring all tasks write to and prioritize this specific memory
• All confirmed translations are automatically saved to this default TM in real-time as you work

TM Configuration Options:

1. Auto-Create (Recommended): The system generates a new TM using your project name and language pair as the default repository
2. Link Existing: Select from your organization's existing TMs to leverage previously translated content
3. Set as Default: Confirm the default status—this setting ensures that every segment translated in this project contributes to this primary memory, even if you access other TMs for reference

Tip: While each task defaults to its assigned TM for writing new translations, you can enable read-access to additional memories during the editing phase for maximum match coverage. However, all new translations are always written to the default TM specified here.

Step 8: Launch Your Project

Review your settings summary:

• Total AI words estimated (based on file size and chosen model multiplier)
• Team assignments (if applicable)
• Deadline settings

Click "Create Project" to generate your translation tasks.

After Creation: What Happens Next

Once created, your project appears in your dashboard with a progress indicator:

• 0%: Pre-processing and AI analysis
• 25%: AI draft generation (if enabled)
• 50%: Human editing phase
• 75%: Review and QA
• 100%: Ready for delivery

Open the project to access the translation editor, where you will see:

• AI-suggested translations in the target column
• Translation Memory matches highlighted in green (drawn from your default TM and any additional read-only memories)
• Term Base reminders popping up for specific words
• Real-time QA checks for missing numbers or double spaces

Tip: If you enabled the Client Portal, your client will automatically see this project listed as "In Progress" when they log in, complete with the percentage complete bar.

Monitoring AI Word Usage

Track your consumption in real-time:

• Each segment translated by AI deducts words from your balance based on the selected model's multiplier
• Translation Memory matches (100% or fuzzy matches) do not consume AI words
• Switching AI models mid-project updates the multiplier for future segments only

If you run out of AI words, your core CAT features remain functional. You can continue translating manually using Translation Memory and Term Base suggestions, or purchase a top-up pack to resume AI drafting.

Troubleshooting Common Issues

"Insufficient AI Words" Error: Reduce the Word Multiplier by selecting a lighter AI model, or enable Translation Memory pre-translation to handle repetitive content without AI costs.

OCR Not Processing: Ensure your scanned file is under the size limit and in a supported image format (PDF, PNG, JPG).

Missing Target Languages: You cannot add new target languages after project creation. Create a new project or duplicate this one for additional languages.

Your first project is now ready. Open the editor to begin refining your AI-generated draft or assigning tasks to your translation team.