Translation Technology: From Fax Machines to AI

Alafranga was founded in 2002, at the end of the fax era. Documents crossed borders by fax, and poor transmission meant retyping by hand. Retyping meant errors, so the translator's first job was often to rebuild the source text before translating it.

 

That is where our attention to the source document began. It still applies today: scans, PDFs and files exported from other systems are checked before any translation starts.

 

Computer-assisted translation tools removed the most wasteful part of the job: translating the same sentence twice. Translation memory stores approved segments and reuses them across a project and across years.

 

We work in memoQ and Trados. Each client has its own translation memory and glossary, built on the first project and maintained on every one after it. The longer the relationship, the faster and more consistent each project becomes. CAT tools also brought accountability: every segment has a history, and every revision is logged.

 

When a large project needs several translators working at the same time, we run it in a shared online CAT environment: Trados or memoQ online, Phrase or Smartcat, depending on the project and on what the client already uses. Everyone on the project works on the same translation memory in real time. A segment approved by one translator is available to the others at once, so nothing is translated twice and terminology stays the same across the team.

 

Your own specialists can take part. Engineers, legal teams and medical staff on your side can be given access to the shared translation memory. When they decide how a term is translated, such as a product name, a regulatory concept or a brand expression, that decision applies to every translator on the project and to every project after it. Project managers see task progress, reviewer assignments and delivery status for every language pair in one view, without status emails.

 

AI is the biggest change since CAT tools, and the most misunderstood. Most AI tools deliver speed. Fewer are open about what happens when the output reads well but is wrong: a misread reference, a term that drifts, a figure that changes.

 

At Alafranga, AI produces a draft and a specialist decides what is delivered. We call this workflow SmartEdit. The draft is produced in four steps:

 

• The whole document, not one sentence at a time. Sentence-by-sentence machine translation loses context: a term can change between pages, and a reference to an earlier section is translated without knowing what it points to. We produce the draft with the full text in view, so a term chosen on page 2 is the same term on page 40.
• The engine is chosen by testing. We work with GPT, Claude and DeepL. Before a programme starts, we run them on your own sample files and use the one that performs best for that content type and language pair. As a rule of thumb, GPT suits general and narrative content, Claude technical and structured content, and DeepL high-fluency European language pairs. The test decides.
• Editing happens in the CAT tool. The draft is imported into memoQ or Trados, checked against your translation memory and glossary, and edited segment by segment by a specialist in your subject.
• A named specialist signs off. Every AI draft is reviewed and approved by a named specialist before delivery. We do not send unedited AI output to clients.

SmartEdit Pro adds an independent second reviewer after editing. It is the route for customer-facing content, regulated material and complex technical documentation. Compare the four workflows

 

On larger or recurring programmes we can also set up AI agents for repeat checks: reading incoming files for problems before translation starts, checking terminology across files, and flagging likely errors before the specialist's review. An agent flags. A person decides.