Guide
Running a multilingual document pipeline: a guide for teams and agencies
Beyond one-off translation
Most translation tools are designed for individual use: paste text, get translation, move on. For teams — law firms with multilingual client bases, immigration agencies processing hundreds of filings, publishers localising for multiple markets — one-off workflows don't scale.
A multilingual team pipeline needs: shared glossaries that everyone uses automatically, per-member access controls, a job queue where work is visible and auditable, and an API for integrating with existing document management systems. This guide explains how to set up and run that workflow in Traxlate.
Organisation accounts
Traxlate's organisation feature lets a Business or Enterprise account add team members who share a credit balance and a unified job history.
Setting up an organisation:
1. Go to Dashboard → Settings → Organisation
2. Create an organisation name
3. Invite members by email — each member receives an invitation link and creates their own Traxlate account that is attached to the organisation
Roles within an organisation:
- Owner: full access, billing, member management
- Admin: can manage members and jobs, cannot manage billing
- Member: can submit and view jobs, cannot see billing or manage members
Member credits come from the organisation pool. The owner can set per-member spending limits to prevent any single member from consuming the entire balance.
Shared glossaries
The most valuable feature for team workflows is the shared glossary.
A glossary is a list of (source term, target term, language pair) tuples that are enforced on every translation run. When a member submits a job, the glossary terms are pinned and appear in the output exactly as specified, regardless of what the translation would otherwise have chosen.
For a law firm translating contracts between English and French:
- "indemnify" → "indemniser" (not "compenser", which is also used in general texts)
- "The Company" → "La Société" (capitalised, as the defined term)
- "force majeure" → "force majeure" (not translated — it is already the technical term in French)
These decisions are made once by the senior translator, stored in the shared glossary, and applied to every document any team member translates without anyone having to think about it.
Creating and managing shared glossaries
From Dashboard → Glossary, click "Create glossary." A glossary has:
- A name (e.g. "Corporate M&A EN→FR")
- A source language
- One or more target languages (the same glossary can cover multiple target languages)
- A list of terms, each with the source phrase and its target equivalent per language
Glossaries are shared at the organisation level. Any member can use any shared glossary. Only admins and owners can modify shared glossaries.
For large term sets (pharmaceutical drug names, legal defined terms), you can bulk-import terms via CSV. The format is: source_term, target_term, language_pair.
API keys for team automation
For automating translation in your existing workflow (e.g. a document management system that automatically translates intake documents), create a dedicated API key for the integration.
From Dashboard → API keys, create a key with:
- A descriptive name (e.g. "DMS integration — intake forms")
- A default glossary (the shared glossary for that document type)
- A monthly credit spend cap (prevents a bug in the integration from draining the balance)
- An optional webhook URL for job completion notifications
Multiple API keys can be created for different integrations, each with their own settings and spend caps. Job history shows which key submitted each job.
Job visibility and audit
For compliance workflows, job history is the audit log.
Every job in Traxlate records:
- Who submitted it (user or API key)
- When it was submitted and when it completed
- The source and target language
- The glossary used
- The credit cost
- The original file name (for document uploads)
Organisation owners and admins can see all jobs submitted by any member, not just their own. This makes it straightforward to answer "which member translated which document" or "what glossary was used on the 15th of March batch."
For export, the job history API (/api/v1/jobs) accepts date-range and member filters, making it suitable for billing reconciliation and compliance reporting.
Translation memory
Translation memory automatically reuses previously translated segments. When a document is submitted, each segment is checked against the memory. If an identical or near-identical segment has been translated before, the stored translation is reused and the credits are not charged.
For organisations that translate batches of similar documents — quarterly reports that share boilerplate sections, recurring immigration forms where the applicant data changes but the instructions do not — translation memory can reduce costs by 30–60%.
Translation memory is shared at the organisation level and builds automatically. No configuration is required; it becomes more valuable over time as more documents are processed.
Webhook integration for document management systems
Most enterprise document management systems support outgoing webhooks (triggering an action when a document is added or modified). Traxlate's API can receive a translation job submission and post a completion notification back to the DMS when the translation is ready.
A typical integration flow:
1. DMS receives an intake document and triggers a webhook to your middleware
2. Middleware calls POST /api/v1/translate with the document
3. Traxlate processes the translation and calls your configured webhook on job completion
4. Middleware receives the completion notification, downloads the translated file, and imports it back into the DMS
This pattern works for law firm document intakes, agency CRM integrations, and publishing localization pipelines.
The completion webhook includes the job ID, status, and a download URL for the translated file. The middleware verifies the X-Traxlate-Signature header before processing.
Volume workflows and batch processing
For processing many documents in a batch (e.g. a discovery production requiring translation of 3,000 exhibits), use the API with a rate-limited submission loop.
The API rate limit is 60 requests per minute per key. For large batches, use a queue in your middleware that submits jobs at 1 request per second and monitors completion via webhook. Do not poll for completion; use webhooks to avoid hammering the jobs endpoint.
For batches larger than 1,000 documents, contact support@traxlate.com — we can allocate dedicated queue priority and provide a dedicated API endpoint with higher rate limits.
Quality assurance in a team workflow
For teams with established quality standards, the recommended workflow is:
1. Train members to review the flagged segments on every job — the accuracy check surfaces exactly where to look
2. Request a human polish pass for documents that carry legal force (immigration filings, court submissions, regulatory filings)
3. Use the translation editor for final review — members can edit the machine translation directly and save the corrected output as the delivery version
The version history in the translation editor records every edit, so the final delivered document has a complete audit trail from machine translation to human review to final delivery.