Documentation

What is SnapIntel?

SnapIntel is a web service for AI translation of DOCX documents and XLSX workbooks. Projects start from native structured-file import with manual language-pair selection.

Instead of copying text into ChatGPT and reformatting the output, SnapIntel gives you a structured workflow:

1. Import a DOCX document directly with manual language selection
2. Analyze the document domain (optional but recommended)
3. Generate and approve a glossary and translation prompt
4. Run translation with the latest ChatGPT model
5. Download formatted delivery DOCX, neutral Excel export, and a QA report

The result is a translation that uses your terminology, follows your instructions, and comes with a quality score plus delivery-ready artifacts.

Quick Start

This is the shortest path from an uploaded DOCX to a completed translation.

Step 1. Create a project

On the Projects page, click New Project. Enter a project name, choose the source and target languages, and upload one or more DOCX files.

You set the source and target languages manually before import starts.

Step 2. Generate a glossary and prompt

Inside the project, click Generate Glossary. SnapIntel reads your document and produces a list of key terms with suggested translations. Review the list and edit any terms that need adjustment.

Then click Generate Prompt. A translation prompt is created based on the document and glossary. You can edit the prompt directly.

Step 3. Save and Approve

Click Save and Approve. This locks in your glossary and prompt. Translation cannot start until this step is completed.

Step 4. Start Translation

Click Start Translation. Progress is shown in real time — by row and by file. You can cancel at any point.

Step 5. Download your results

When translation completes, download your files from the Results section: formatted delivery DOCX, neutral Excel export, and QA PDF report.

Plans and Quotas

SnapIntel has three plans:

How quotas work

Word count is calculated from the source text in your uploaded project files and is deducted at the point of import — not at the point of translation. If a document exceeds your per-document limit or would push you over your monthly quota, the project is still created but translation is blocked until you upgrade or your quota resets.

Monthly quotas reset at the start of each billing cycle. The current quota usage is visible in Settings → Plan & Usage.

Agency plan — unlimited words

The Agency plan has no word limits because you provide your own OpenAI API key. You pay OpenAI directly for each translation job. SnapIntel routes the job through your key without adding markup. See the BYOK section for setup instructions.

Step 1 — Import

SnapIntel uses one explicit import mode: native DOCX/XLSX intake with manual language-pair selection.

What to upload

• File formats: .docx or .xlsx
• Upload the original document and choose the language pair manually
• Multi-file projects require one shared language pair across all files

Multi-file projects

You can upload multiple files in one project. All files must share the same source and target language pair. Each file is translated independently, and results are downloaded per file.

What happens during import

SnapIntel extracts translatable content, builds the translation input, and keeps the original file for delivery reconstruction.

Step 2 — Domain Analysis (Optional)

Domain analysis is an optional first step that reads your document and identifies its subject area, key topics, and terminology context. This information is then used when generating the glossary and translation prompt.

When to use it

Domain analysis is most useful for specialized documents — technical manuals, legal contracts, financial reports, engineering specifications. For general-purpose text, you can skip it and go straight to glossary generation.

What it produces

• Primary domain (e.g. Mining, Legal, Finance, Energy, Engineering)
• List of main subject matters
• Short summary of document content

The domain analysis result is shown in the project and is used automatically when you click Generate Glossary or Generate Prompt. You cannot edit the domain analysis output directly — but you can edit the glossary and prompt it produces.

Step 3 — Glossary

The glossary tells the AI exactly how to translate specific terms. Each line is a rule: translate this term as this other term.

Glossary format

Each line must follow this exact format:

Translate "source term" as "target term"

Example:

Translate "ball mill" as "шаровая мельница"
Translate "overburden" as "вскрышные породы"
Translate "throughput" as "производительность"

No other format is supported. Headers, comments, numbered lists, or tables will not be interpreted correctly.

Generating a glossary

Click Generate Glossary. SnapIntel extracts key terms from your document and suggests translations. If you ran domain analysis first, the glossary will reflect the identified domain and terminology context.

Editing the glossary

You can edit any line, add new terms, or delete terms you do not need. You can also clear the auto-generated glossary entirely and paste your own.

Using your own glossary

If you have an existing terminology list, paste it into the glossary field in the correct format. Make sure each line follows the Translate "X" as "Y" pattern.

Step 4 — Translation Prompt

The translation prompt is the instruction set given to the AI for every translation batch. It defines tone, style, and any document-specific rules.

Generating a prompt

Click Generate Prompt. A prompt is generated based on your document, domain analysis (if run), and glossary. The prompt is fully editable.

Editing the prompt

You can modify any part of the prompt. Common additions:

• Specify register or formality level
• Request preservation of specific formatting patterns
• Add client-specific instructions
• Specify how to handle untranslatable terms (leave in source language, transliterate, etc.)

Step 5 — Save and Approve

Save and Approve is the gate between preparation and translation. It locks in your glossary and prompt and signals that the AI can proceed.

Why this step exists

Translation cannot start until you explicitly approve. This prevents the AI from running on an unreviewed glossary or a default prompt you did not intend to use.

What triggers re-approval

• Any edit to the glossary
• Any edit to the translation prompt

If you edit either field after approving, the approval resets and you must click Save and Approve again before translation can start.

Step 6 — Start Translation

Click Start Translation. The job is queued and begins processing immediately.

Progress tracking

Translation progress is shown in real time:

• Rows translated vs. total rows
• Current batch number
• Per-file progress for multi-file projects
• Recent workflow events log

Batch processing

Translation runs in batches — groups of rows sent to the AI in a single request. This is handled automatically. You do not need to configure batch size.

Cancellation

You can cancel a running translation at any point by clicking Cancel. The job stops after the current batch finishes. Rows already translated are preserved and available for download as a partial result.

Re-running translation

You can start translation again on the same project after cancellation or failure. The job will re-use the existing approved glossary and prompt. Previous results will be overwritten.

Step 7 — Results and Downloads

When translation completes, the Results section shows a summary: rows translated, quality score, and download links.

Available downloads

• Translated DOCX — the translated document rebuilt into the original file structure
• Excel (TM export) — neutral two-column source/target file for TM import in CAT tools
• QA PDF — full quality report with score, executive summary, and per-segment findings

Using Excel in TM workflows

To use the translated segments in a translation memory workflow:

1. Download the Excel file from Results
2. Open the translation memory import flow in your CAT tool
3. Map the source and target columns if the tool requires explicit column mapping
4. Import the Excel file into your CAT tool or translation memory workflow as a neutral source/target spreadsheet

Translated DOCX

The translated DOCX is your delivery file. SnapIntel rebuilds the translated content back into the original document structure.

Use it to review the translation, share it with a client, or deliver it directly.

Neutral Excel Export

The Excel file is a neutral two-column source/target spreadsheet. It is designed for TM import workflows across CAT tools rather than for one vendor only.

How to use it

1. Download the Excel file from the project Results
2. Open the TM import flow in your CAT tool
3. Map source and target columns if your tool requires it
4. Import and apply your overwrite policy

QA Report

Every completed translation job includes an automated quality assessment.

What the report contains

• Overall quality score — a rating from 0 to 10
• Executive summary — a short plain-language description of the translation quality
• Top strengths — what the AI did well
• Top risks — main issues identified
• Category breakdown — scores across five dimensions: Accuracy, Terminology, Fluency, Grammar, Cultural Fit
• Flagged segments — individual segments with issues, classified as HIGH, MEDIUM, or LOW severity, with QA notes

Using the report

The QA report gives you a basis for deciding whether the translation needs post-editing before delivery. HIGH-severity findings point to segments that likely need manual review. The PDF version can be shared with a client as evidence of quality control.

What is BYOK?

BYOK (Bring Your Own Key) means you connect your own OpenAI API key to SnapIntel. Instead of using SnapIntel's shared platform key, all translation jobs in your account are routed through your key.

This is the only option on the Agency plan — and it comes with a key benefit: no word limits. Because you pay OpenAI directly for usage, SnapIntel does not impose a monthly word cap.

What SnapIntel does with your key

• Your key is stored encrypted in the database and is never returned to the UI in plaintext
• It is used only to send translation requests on your behalf
• It is not used for any other purpose

How to Get and Configure Your OpenAI API Key

Step 1. Create an OpenAI account

Go to platform.openai.com and sign up or log in. You need an OpenAI account with billing enabled to generate an API key that can process requests.

Step 2. Add a payment method in OpenAI

In your OpenAI account, go to Settings → Billing and add a credit card. OpenAI charges you directly for API usage based on tokens processed. Translation jobs in SnapIntel use the latest available GPT model — check OpenAI's pricing page for current rates.

Step 3. Generate an API key

1. In your OpenAI account, go to API keys (platform.openai.com/api-keys)
2. Click Create new secret key
3. Give it a name — for example, "SnapIntel"
4. Copy the key immediately — OpenAI shows it only once

Step 4. Add the key in SnapIntel

1. In SnapIntel, go to Settings
2. Scroll to OpenAI API Key
3. Paste your key into the field (starts with sk-...)
4. Click Save

The key is saved and encrypted. The field will show a masked version confirming it is configured. Translation jobs on your account will now use your key.

Step 5. Verify it works

Create a project, go through the workflow, and start a translation. If your key is valid and billing is active in OpenAI, the job will run normally. If the key is invalid or billing is not set up, the translation will fail with an error from OpenAI.

Replacing or removing your key

To replace your key, paste a new one into the API Key field in Settings and click Save. To remove it, click Clear. Note that removing your key will block translation until a new key is saved.

Import Errors

Import fails with a validation error

Import can fail if the document contains no supported visible content after hidden-text filtering or if the file is malformed.

Fix: Retry with a standard .docx document and review any unsupported-content warnings.

Language pair not detected

SnapIntel does not auto-detect the language pair during intake. You choose the source and target languages manually before import starts.

Fix: Re-run the import and make sure the correct source and target languages are selected.

File uploads but project shows as failed

The file may have passed initial validation but failed during normalization. This can happen with malformed DOCX XML or unsupported structures that leave no translatable supported blocks.

Fix: Retry with the original DOCX file. If the issue persists, contact us at hello@snapintel.io with the project name.

Translation Blocked or Quota Issues

"Start Translation" button is disabled

Translation requires an approved glossary and prompt. If you have not clicked Save and Approve, or if you edited the glossary or prompt after approving, the button will be disabled.

Fix: Click Save and Approve.

Translation blocked — quota exceeded

Your monthly word quota or per-document limit has been reached. The project was created and the document is prepared, but translation jobs cannot start until the quota is available.

Fix: Upgrade your plan, or wait for your monthly quota to reset. Quota usage is shown in Settings → Plan & Usage.

Translation blocked — Agency plan requires OpenAI key

The Agency plan requires a saved OpenAI API key. If no key is configured, all translation jobs are blocked.

Fix: Go to Settings → OpenAI API Key and save your key. See the BYOK setup guide for instructions.

Plan is suspended

If a payment fails and is not resolved within 7 days, the plan is suspended and translation is blocked.

Fix: Update your billing details through Paddle's billing portal or contact hello@snapintel.io.

Glossary and Prompt Issues

Glossary generation returns an empty result

This can happen with very short documents or documents with very little specialized terminology.

Fix: You do not need a glossary to run translation. Leave it empty or paste your own terms in the correct format, then proceed to prompt generation.

Terminology in the translation does not match the glossary

The glossary is provided as instructions to the AI, not as a hard replacement rule. The AI may occasionally deviate, especially for terms that appear in ambiguous contexts.

Fix: Make sure the glossary entries follow the exact format: Translate "X" as "Y". Entries in any other format are not interpreted as instructions. For critical terms, you can reinforce them in the translation prompt as well.

Save and Approve resets after editing

This is expected behavior. Any edit to the glossary or prompt resets approval. You must click Save and Approve again before starting translation.

Results and Download Issues

Translation completed but some rows are missing

If the job was cancelled or failed partway through, the results will be partial. The Results section shows how many rows were translated vs. the total.

Fix: Start translation again — the job will re-run on the full document. Previous results will be overwritten.

Download links are not available

Downloads appear only after translation has completed successfully. If the project is still in progress or in a failed state, the download links will not be shown.

Fix: Wait for the job to complete, or restart translation if it failed.

QA PDF is not available

QA PDF generation runs after translation completes. On rare occasions it may take an additional minute to appear.

Fix: Refresh the page. If the QA PDF is still missing after a few minutes, contact hello@snapintel.io.

Excel import into a CAT TM fails

The Excel export is neutral, but each CAT tool still applies its own TM import rules. Imports can fail if the target TM uses a different language pair, requires explicit source/target column mapping, or expects a vendor-specific import profile.

Fix: Make sure the target TM has the same language pair as your translation project and map the source/target columns correctly. If a specific CAT tool still rejects the file, review that tool's TM import requirements or contact its support team with the error message.

General

Do I need a CAT-tool account to use SnapIntel?

No. Projects start from native DOCX/XLSX import. A separate CAT-tool account is only needed if you want to import the neutral Excel export into your own downstream TM workflow.

Can I use SnapIntel without domain analysis?

Yes. Domain analysis is optional. You can go directly from import to glossary generation, or even skip the glossary and go straight to prompt generation and translation.

Can I translate the same project more than once?

Yes. You can start a new translation run on the same project at any time. Each run uses the current approved glossary and prompt. Previous results are overwritten.

What languages are supported?

You choose the source and target languages manually before import starts. In practice, any pair that OpenAI's models can translate is a candidate workflow.

Is my data used to train AI models?

No. Your document content is not used for AI training. Documents are processed through OpenAI's API under our business agreement, which includes data processing terms. See the Privacy Policy for full details.

Where is my data stored?

All data is stored on servers in the European Union. Document data is deleted when you delete a project.

Billing

How are payments processed?

Payments are processed by Paddle, the Merchant of Record for all SnapIntel subscriptions. SnapIntel never stores your card details. All billing, invoicing, and subscription management is handled by Paddle.

How do I cancel my subscription?

You can manage or cancel your subscription through Paddle's billing portal. Access it from your account settings or contact hello@snapintel.io.

What happens to my data if I cancel?

Your projects and files remain accessible until you delete them or close your account. Cancelling a paid plan downgrades you to the Free plan — you do not lose access to the application.

How do refunds work?

Refunds are handled in accordance with Paddle's refund policy. To request a refund, contact hello@snapintel.io or Paddle's billing support directly.