Automator Guide¶

You're a power user. In the 2025 survey, 52.1% of MT users translate at least weekly — with 31% translating more than once per week. You need batch processing, translation memory, and workflow integration.

This guide covers advanced features for high-volume translation work.

Your Profile¶

From the survey:

Statistic	Your Needs
31% translate more than once/week	High-volume workflows
59.9% use MT for written communication	Email, documents, reports
56.7% use MT to read/understand content	Research, analysis
42% use MT for publishing	External communications
76.5% work on laptops	Desktop-first workflows

You need more than one-off translations. You need a system.

Batch Processing¶

Translate Entire Folders¶

Instead of one document at a time:

from pauhu import Pauhu

client = Pauhu()

# Translate all documents in a folder
results = client.batch.translate(
    source_folder="/documents/incoming/",
    target_folder="/documents/translated/",
    target_language="fi",
    preserve_format=True
)

# Results
print(f"Translated: {results.success_count} documents")
print(f"Total words: {results.word_count}")
print(f"Time: {results.elapsed_time}")

CLI for Bulk Operations¶

# Translate all PDFs in a directory
pauhu translate ./input/*.pdf --target fi --output ./output/

# Watch a folder for new documents
pauhu watch ./inbox/ --target fi --output ./translated/

# Process with specific domain model
pauhu translate ./legal-docs/ --target fi --domain "12 Law"

Supported Formats¶

Format	Extensions	Features
Office	.docx, .xlsx, .pptx	Full formatting preserved
PDF	.pdf	OCR if needed, layout preserved
Text	.txt, .md, .html	Clean conversion
Data	.json, .xml, .csv	Structure preserved
Images	.png, .jpg, .tiff	OCR extraction
Scans	.pdf (scanned)	OCR + layout analysis

File Hubs (Unlimited Projects)¶

Organize your work. Each project gets its own:

Translation memory
Terminology glossaries
Quality settings
Team access controls

Create a Project¶

# Create a new file hub
project = client.projects.create(
    name="EU Regulation Q1 2025",
    domain="10 European Union",
    languages=["en", "fi", "sv"],
    glossary="./eu-terms.csv"
)

# Upload documents
for doc in regulatory_documents:
    project.upload(doc)

# Translate entire project
results = project.translate(target="fi")

Project Structure¶

📁 EU Regulation Q1 2025
├── 📁 Source Documents
│   ├── regulation-2025-001.pdf
│   ├── directive-ai-act.docx
│   └── council-conclusions.pdf
├── 📁 Translations (Finnish)
│   ├── regulation-2025-001_fi.pdf
│   ├── directive-ai-act_fi.docx
│   └── council-conclusions_fi.pdf
├── 📁 Translation Memory
│   └── project-tm.tmx
├── 📁 Glossaries
│   └── eu-terms.csv
└── 📁 Quality Reports
    └── qa-report.html

Translation Memory¶

Reuse previous translations. 100% matches are instant and included.

How It Works¶

graph LR
    A[New Segment] --> B{In TM?}
    B -->|100% match| C[Use Previous]
    B -->|Fuzzy match| D[Suggest + Translate]
    B -->|No match| E[New Translation]
    C --> F[Instant]
    D --> G[Discounted]
    E --> H[Full Cost]

TM Statistics¶

From typical enterprise usage:

Match Type	Percentage	Cost
100% match	15-40%	Included
95-99% fuzzy	10-20%	50% cost
75-94% fuzzy	10-15%	75% cost
New	35-65%	Full cost

Average savings: 25-45% on repeat content.

Import Existing TM¶

# Import from other CAT tools
project.tm.import_file(
    "./existing-tm.tmx",  # TMX format
    # Or: "./legacy-tm.xliff"  # XLIFF format
    # Or: "./sdl-tm.sdltm"  # SDL Trados
)

# Export for backup
project.tm.export("./backup-tm.tmx")

Auto-Recognition¶

Drop files — Pauhu handles the rest.

What Gets Detected¶

Aspect	Detection	Accuracy
Source language	Automatic	99%+
Document domain	21 EuroVoc domains	95%+
Document type	Contract, letter, form, etc.	90%+
Terminology	Domain-specific glossary	Automatic

Domain Routing¶

# Upload any document — domain detected automatically
doc = project.upload("./unknown-document.pdf")

print(doc.detected_domain)      # "12 Law"
print(doc.detected_type)        # "Contract"
print(doc.detected_language)    # "en"
print(doc.terminology_applied)  # ["legal-fi", "contract-terms"]

Quality Assurance¶

Multi-Tier QA¶

Every translation goes through:

graph TB
    A[Translation] --> B[Terminology Check]
    B --> C[Consistency Check]
    C --> D[Style Check]
    D --> E[Quality Score]
    E --> F{Score > 85?}
    F -->|Yes| G[Auto-Approve]
    F -->|No| H[Human Review Queue]

QA Configuration¶

# Set project-level QA rules
project.qa.configure(
    terminology_strict=True,      # Enforce glossary terms
    consistency_threshold=0.9,    # Flag inconsistent translations
    style_guide="formal",         # Formal/informal/technical
    auto_approve_threshold=85,    # Minimum score for auto-approval
    forbidden_terms=["TBD", "TODO", "FIXME"]
)

QA Report¶

┌─────────────────────────────────────────────────────────┐
│                    QA REPORT                            │
│                 Project: EU Reg Q1 2025                 │
├─────────────────────────────────────────────────────────┤
│ Documents processed: 47                                 │
│ Total segments: 12,456                                  │
│ Average quality score: 92.3                             │
│                                                         │
│ ISSUES FOUND:                                           │
│ ├── Terminology: 23 (glossary mismatches)              │
│ ├── Consistency: 12 (same source, different target)     │
│ ├── Style: 8 (informal register in formal doc)         │
│ └── Numbers: 3 (possible format errors)                 │
│                                                         │
│ RECOMMENDATIONS:                                        │
│ • Review segments flagged for terminology               │
│ • Update glossary with 5 new terms detected             │
│ • Consider post-editing for 8 style issues             │
└─────────────────────────────────────────────────────────┘

API Integration¶

RESTful API¶

import requests

# Translate via API
response = requests.post(
    "https://api.pauhu.ai/v1/translate",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "text": "EU directive on artificial intelligence",
        "source": "en",
        "target": "fi",
        "domain": "10 European Union"
    }
)

result = response.json()
print(result["translation"])  # "EU:n tekoälydirektiiviä"
print(result["quality_score"])  # 94

Webhook Integration¶

# Configure webhook for batch completion
client.webhooks.create(
    url="https://your-system.com/pauhu-webhook",
    events=["batch.completed", "document.translated"],
    secret="your-webhook-secret"
)

CMS/DAM Integration¶

System	Integration	Status
SharePoint	Native plugin	✅ Available
Confluence	REST API	✅ Available
WordPress	Plugin	✅ Available
Contentful	Webhook	✅ Available
Custom	API + Webhooks	✅ Full support

Workflow Automation¶

Watch Folders¶

# Auto-translate documents dropped in folder
pauhu watch ./inbox \
    --target fi \
    --output ./translated \
    --notify email:team@org.fi \
    --qa-threshold 85

Scheduled Jobs¶

# pauhu-schedule.yaml
schedules:
  - name: "Daily EU News"
    source: "https://europa.eu/newsroom/rss"
    target: "fi"
    schedule: "0 6 * * *"  # 6 AM daily
    output: "./news/translated/"

  - name: "Weekly Reports"
    source: "./pending-reports/"
    target: "fi,sv"
    schedule: "0 8 * * 1"  # Monday 8 AM
    output: "./reports/translated/"

Integration with Your Tools¶

# Example: SharePoint integration
from pauhu.integrations import SharePointConnector

sp = SharePointConnector(
    site="https://yourorg.sharepoint.com/sites/documents",
    credentials=credentials
)

# Watch SharePoint folder, translate, upload back
sp.watch_folder(
    source_folder="/Incoming/",
    target_folder="/Translated/",
    target_language="fi"
)

Cost Management¶

Usage Dashboard¶

┌─────────────────────────────────────────────────────────┐
│                 USAGE THIS MONTH                        │
├─────────────────────────────────────────────────────────┤
│ Characters translated: 2,456,789                        │
│ Documents processed: 234                                │
│ TM savings: 892,345 characters (36%)                   │
│                                                         │
│ COST BREAKDOWN:                                         │
│ ├── New translations: €412.50                          │
│ ├── Fuzzy matches: €89.20                              │
│ ├── 100% matches: €0.00                                │
│ └── Total: €501.70 (€0.20/1000 chars effective)        │
│                                                         │
│ Budget remaining: €498.30 / €1,000.00                  │
└─────────────────────────────────────────────────────────┘

Cost Optimization Tips¶

Build TM early — First project costs more, subsequent projects cost less
Use consistent terminology — Reduces fuzzy match variations
Batch similar documents — Better TM leverage
Pre-translate with TM — 100% matches are included

Team Features¶

Role-Based Access¶

# Set up team with different access levels
project.team.add_member("translator@org.fi", role="translator")
project.team.add_member("reviewer@org.fi", role="reviewer")
project.team.add_member("admin@org.fi", role="admin")

Workflow Stages¶

graph LR
    A[Upload] --> B[Auto-Translate]
    B --> C[QA Check]
    C --> D{QA Pass?}
    D -->|Yes| E[Auto-Approve]
    D -->|No| F[Human Review]
    F --> G[Post-Edit]
    G --> H[Final Approval]
    E --> I[Published]
    H --> I

Next Steps¶

API Reference

Full API documentation for developers.

API Docs
Translation Memory

Deep dive into TM features.

TM Guide
Quality Assurance

Configure QA rules and workflows.

QA Guide
Enterprise Deployment

On-premises, air-gapped, or hybrid.

Deployment

Data source: Koponen, Nurminen & Ilkiliç (2025). "Automaattisten käännössovellusten käyttö Suomen julkishallinnossa." University of Eastern Finland. URN:ISBN:978-952-61-5851-8