Skip to content



AI Translation Workflow Consulting

Audit, Build and Run: from the first decision to day-to-day operation


We have worked on large-scale machine translation post-editing projects as part of collaborative teams since 2017, starting with 2.2 million words of Amazon and Alibaba product content for Jonckers. In 2018 we built SmartEdit, our controlled AI-assisted translation workflow, and we have used it in client delivery ever since. We know where AI drafting works, where it fails, and where the quality gates have to sit.


We offer that experience to companies designing their first AI translation workflow, to LSPs rebuilding their production infrastructure, and to teams that already use AI translation but cannot rely on the output.

 
 
▮ Three packages. Most clients start with Audit.
Each package can be bought on its own. Together they follow the order in which this work usually happens: decide, build, operate.
 

▮AUDIT: Decide what goes to AI and what stays human

Not every company needs an audit. If your in-house AI translation is working and you can show how it is checked, there is nothing to fix. If you are not sure, or the gaps are already showing, we assess where you stand and set out the controls that fit your content, your tools and your level of risk, in a written report and a workflow designed for your organisation.
.
  • Content classification: every content type you produce, sorted into AI draft with review, AI draft with dual review, or fully human translation
  • Risk mapping: where an error would have legal, safety or commercial consequences
  • Quality gate design: what is checked, by whom, and at which step
  • Asset review: the condition of your translation memories, glossaries and current tools
  • Cost comparison: your current workflow against the proposed one
  • You receive: a written workflow specification and roadmap that your team can implement and your clients can audit.
  • Pricing: fixed fee, agreed before work starts. Typical duration two to four weeks.

▮BUILD: Set up the workflow and the systems around it 

Once the decisions are made, the workflow has to work inside real tools, with real translators, on real deadlines. We build it in your environment, not ours.
  • AI drafting inside your CAT tool: GPT, Claude or DeepL connected via API to memoQ, Trados or Smartcat, with your TM and glossary applied to every segment
  • Prompt design: instructions that keep AI output consistent with your terminology and style
  • Automated QA: checks for numbers, tags, placeholders and terminology drift before anything reaches a reviewer
  • Platform setup: client portals, vendor onboarding and assignment, shared TM across language pairs
  • Migration: moving off legacy tools and fragmented TMs, with TM cleaning, consolidation and a clear account of what the move will cost in match quality
  • Content system connections: CMS and XLIFF pipelines, API routing by language pair or content type. Integration with ERP and other document systems is scoped case by case.
  • You receive: a working, tested workflow, full technical documentation, and a briefing for your team.
  • Pricing: fixed project fee against a written scope. Changes outside the scope are quoted separately before work on them starts.

▮RUN: We operate the workflow for you

For companies that produce content continuously and do not want to build an in-house language team. We run the workflow, keep the language assets in order, and report on quality every month.
  • Production: AI drafting and human review across all your languages, with a named reviewer per language
  • Asset maintenance: TM and terminology kept current as your products change
  • Quality reporting: a monthly report on error types, review findings and terminology changes
  • One point of contact for every language and content type
  • Pricing: a monthly retainer with an agreed volume band, a fixed price per product release, or a combination of both. No per-word invoicing.
▮What we do not send to AI
Medical device documentation, texts with legal effect, and safety-critical instructions run through HumanExpert, our fully human workflow, by default. Where a sworn translator's signature is required, the document stays with a sworn translator. An audit that recommends AI for everything is not an audit.
 


Our Linguist Network 
484 active specialists with real project history. 5,200 registered candidates across 40+ language pairs. Six-stage application process since 2009.

→ Global Vendor Network