Is this for freelancers or in-house teams, or both together?

Freelance and in-house translators, localization project managers and coordinators, terminology and quality leads, transcreation specialists, and language-service provider production teams, most often together. Cohorts can split by role when the exercises need to go deeper for one group than another.

Does anyone need to code?

No coding or data-science background is expected, just familiarity with your language pairs, your translation workflow, and the CAT tools you already run.

How long does it run, and can it happen online?

The core program is two days, about 13 hours in total, delivered onsite or live online. It splits into half-day sessions when project deadlines or translator schedules need it to.

Will the syllabus match our language pairs and CAT stack?

Before the training we review your language pairs, your CAT and translation-management tools, client requirements, and the project types you handle most. Exercises and templates get built around that, without needing live confidential client content in the room.

How is pricing worked out for translator training?

Group size, delivery format, customization depth, and location set the price, and every language service provider gets its own scope. Send us your team structure and goals and we'll put together a proposal that reflects them.

What do participants walk away with?

Everyone keeps the prompt library built during the sessions. Glossary templates and QA checklists go with them too. So does the pilot plan drafted on the last day. We stay reachable for follow-up questions afterward, and an optional refresher or advanced session can be scheduled once adoption is underway.

Which AI and CAT tools get used?

Core practice happens on mainstream MT engines and assistants like ChatGPT, Microsoft Copilot, and Gemini, then carries over to whatever CAT and translation-management tools your organization runs. Raw MT doesn't replace a trained linguist here, and nothing in the syllabus assumes it should publish without review.

How is client and source content kept safe in the room?

No live client file is used. Source texts are written for the course, and where one began as a client job it has been fully anonymized. Participants practise classifying source and client content while using only approved tools. They send an AI system no more than the job requires. A human reviews every translated, post-edited, or transcreated deliverable before a client sees it, every time.

AI Training · Translators & Localization

AI Training for Translators & Localization Teams

Machine translation drafts nearly everything now. The question is whether that draft is actually right for the reader in front of you, and that call still belongs to a trained linguist.

This training builds the skills around that call: post-editing efficiently, holding terminology steady across a project, running QA that catches what actually matters, and drafting transcreation options faster. The model never gets the final word on what "correct" means.

modules: 6
hours: 13

Some of the 500+ brands we've worked with

Selected brands we’ve worked with across industries.

See all references

Machine-translation post-editing means correcting an MT draft against the source until it meets an agreed quality bar, light or full, before a client sees it. Localization stretches that discipline across an entire project. The glossary stays consistent while QA checks names and numbers against source. Someone owns client and vendor communication. Generative AI now touches every one of those steps.

A fluent mistranslation or a drifting term still slips past a casual read, and pasting a confidential document into an ungoverned tool turns a quality risk into an exposure one. AI handles the volume — comparing drafts, flagging segments, drafting glossary and transcreation candidates — but never judges fit or tone; that call stays with the linguist. Editorial work past translation lives in AI training for media and publishing, with the full set under AI training.

A fluent draft can still be wrong

Modern MT/LLM output reads smoothly enough that errors slip past a casual read — training builds a structured check against the source.

One wrong term, repeated across four hundred files

One inconsistent term erodes a whole multi-file project — AI drafts glossary candidates, the terminology lead still approves entries.

Deadlines don't leave much room for a real check

Deadlines compress QA time — AI flags suspicious numbers, names, and formatting against source before a client ever sees them.

Can AI write a tagline that lands in another culture?

Marketing copy rarely survives a literal translation — AI drafts adapted starting points, the transcreator weighs brand fit before it ships.

Email eats billable hours

Coordinators lose hours to client queries and status updates — AI drafts them faster, the sender still verifies scope and commitments.

01
How machine translation and language models differ
120 minBeginner
MT engines and general-purpose language models generate translations in different ways, and that difference shows up in where the output is reliable and where a trained linguist stays essential. Participants sort sample segments by risk and required effort before touching a live project.
How MT differs from a language model
Where a person still has to decide
Confident errors: hallucinations and mistranslations
Quality tiers, named
One client review standard
02
A repeatable post-editing workflow
150 minIntermediate
A workflow that compares MT output against the source, fixes what's genuinely broken, and respects the agreed quality level, all without rewriting every sentence from scratch.
Comparing MT output against the source
Light edit or full edit?
Fixing terminology and grammar, with the right register
When a segment needs retranslation
Tracking edit effort for better estimates
03
Turning scattered preferences into one term base
120 minIntermediate
Client preferences and past projects rarely live in one place until someone builds a term base from them. AI speeds up candidate extraction. Approval and ownership stay with the terminology lead.
Building a glossary that survives turnover
AI drafts, human approves
Client rules versus industry standard usage
Catching drift across a big project
Wiring glossaries into CAT and AI tools
04
QA that catches what a client would
120 minIntermediate
A structured pass finds the errors that matter most: omissions, additions, mismatched numbers and names, and tone that drifts across a document set. AI runs the scan, and the linguist confirms every flagged issue before delivery.
Scanning for omissions and additions first
Checking numbers, names, dates, units, formatting
Holding tone and register steady throughout
AI flags, a person verifies
A QA checklist tied to client rules
05
From literal translation to something that lands
150 minIntermediate
Marketing and brand content rarely survives a word-for-word pass. Participants draft several culturally adapted variants and test them against the target market. They present them with a rationale a reviewer signs off on before anything goes live.
Where translation ends and transcreation starts
Adapting culture without losing the brand
Testing wordplay and idiom against the market
Presenting options with a clear rationale
Keeping brand and glossary rules in the mix
06
Project communication and where to draw lines
120 minAdvanced
Boundaries for source and client content, an approved AI-and-CAT toolset, and a human checkpoint mapped into every stage, so the group leaves with a pilot plan tied to quality and turnaround.
Drafting client and vendor updates fast
Sorting source content by confidentiality
Approved AI and CAT tool pairings
Human checkpoints in MTPE, QA, transcreation
Who owns quality and delivery through sign-off
Picking pilots with a measurable target

What you will learn

Know whether raw MT needs a light or full edit
Post-edit efficiently, catching mistranslations and register slips
A term base that lasts
Run QA passes that catch omissions and additions, including mismatches
Drafts that keep brand intent
Clear client and vendor communication about scope and timelines
Classify content, then follow the tool and data rules
Human review at every MTPE, QA, or transcreation step

Who should attend

Freelance and in-house translators
Localization project managers and coordinators
Whoever owns terminology and language quality
Transcreation and marketing-localization specialists
Language-service provider production and vendor-management people
In-country reviewers checking the final language call

Translators spend two days in hands-on sessions built around realistic translation and localization work. The schedule splits cleanly into half days, and we adapt the exercises to your language pairs, CAT tools, and client requirements.

Format: Onsite or live online
Duration: 2 days (about 13 hours, can be split into half-day sessions)
Group size: Up to 20 participants per group
Language: English or Turkish
Materials: Prompt library, QA checklists, and practice scenarios
Certificate: Certificate of completion

Using AI in Content Teams6 modules · 13 hoursAI training for content and editorial teams, scoped to your actual calendar: briefs, drafting, repurposing, and a house style that holds up across freelancers and every format derived from the piece.View training

AI Literacy & Prompt Engineering6 modules · 13 hoursAI literacy and prompt engineering training for your whole company. Teams practice on their own workflows across ChatGPT, Copilot, and Gemini, and walk out with a shared prompt library they use again the next day.View training

AI in Media & Publishing6 modules · 13 hoursNewsrooms, publishers, and editorial teams get hands-on generative AI training: research and transcript summaries, archive retrieval, metadata, localization, repurposing, and accountable publishing workflows.View training

AI Training for Newsrooms6 modules · 13 hoursAI training for journalists and newsrooms: reporting research, interview transcription, verification support, editorial drafting, source protection, and correction and disclosure practice, built around a reporter's workflow.View training

You'll leave with a prompt library and QA checklists, plus a pilot plan built around your actual projects. To get there we need your language pairs and the CAT tools already in your stack. Show us where post-editing or QA slow things down most.

Two colleagues handing over a reviewed work item between their role stations

Is this for freelancers or in-house teams, or both together?
Freelance and in-house translators, localization project managers and coordinators, terminology and quality leads, transcreation specialists, and language-service provider production teams, most often together. Cohorts can split by role when the exercises need to go deeper for one group than another.
Does anyone need to code?
No coding or data-science background is expected, just familiarity with your language pairs, your translation workflow, and the CAT tools you already run.
How long does it run, and can it happen online?
The core program is two days, about 13 hours in total, delivered onsite or live online. It splits into half-day sessions when project deadlines or translator schedules need it to.
Will the syllabus match our language pairs and CAT stack?
Before the training we review your language pairs, your CAT and translation-management tools, client requirements, and the project types you handle most. Exercises and templates get built around that, without needing live confidential client content in the room.
How is pricing worked out for translator training?
Group size, delivery format, customization depth, and location set the price, and every language service provider gets its own scope. Send us your team structure and goals and we'll put together a proposal that reflects them.
What do participants walk away with?
Everyone keeps the prompt library built during the sessions. Glossary templates and QA checklists go with them too. So does the pilot plan drafted on the last day. We stay reachable for follow-up questions afterward, and an optional refresher or advanced session can be scheduled once adoption is underway.
Which AI and CAT tools get used?
Core practice happens on mainstream MT engines and assistants like ChatGPT, Microsoft Copilot, and Gemini, then carries over to whatever CAT and translation-management tools your organization runs. Raw MT doesn't replace a trained linguist here, and nothing in the syllabus assumes it should publish without review.
How is client and source content kept safe in the room?
No live client file is used. Source texts are written for the course, and where one began as a client job it has been fully anonymized. Participants practise classifying source and client content while using only approved tools. They send an AI system no more than the job requires. A human reviews every translated, post-edited, or transcreated deliverable before a client sees it, every time.

By team

By industry

By tool

AI Training for Translators & Localization Teams