Exception Review

Qualified human decisions on the segments your pipeline flags.

--:--:-- [PST]
S01

What it is

Your engine translates. A model scores every segment for how likely it is to be wrong. An automated editing pass resolves what it can. Only what is left reaches a person.

That residue is what we work on — and by definition it is the hard part. Ambiguity the model could not resolve. Terminology that has to match a document nobody gave the machine. Formality and register the source language never stated. Content where being confidently wrong is expensive.

S02

What comes back

01
Vanenberg
Accepted
The flag was wrong. Publish it.
02
Vanenberg
Corrected
The machine's version, edited into shape.
03
Vanenberg
Retranslated
The machine's version could not be salvaged, so we translated the segment from source.
04
Vanenberg
Escalated
Not a language problem. Wrong source, missing context, a term base that contradicts itself.
S03

Every disposition carries a named reviewer and a timestamp

That record is part of the deliverable, not an internal artefact — in regulated content it is often the part your own client is buying.

S04

Where translation comes in

Where the machine's version cannot be edited into shape, we translate the segment properly. It happens more often in some languages than others: in the 2025 evaluation, professional human translation beat every machine system tested into Japanese, Icelandic, Estonian and Egyptian Arabic. Those are the pairs where retranslated is regularly the correct disposition, and where a per-word post-editing discount is hardest to justify. More on that under human translation.

S05

How we work with your automated checks

We assume they exist. Buyers increasingly run quality estimation, an automated review pass, or both, and route only what falls below a threshold to a person. That is a sensible design and we are built to sit at the end of it. Send us the exceptions.

Where it matters, we will also tell you when we think a threshold is set wrong — which is a conversation the party that configured the threshold cannot easily have with you.

S06

Independence

We do not sell a translation engine. We do not sell a platform. On the work we review, we were not the producer. An assessment written by the party that produced the output is weaker evidence, and regulated buyers have started saying so out loud.

Where we have produced the content ourselves, we say so and we do not also certify it.

S07

What it costs

Per segment, or per hour. Not per word — per-word pricing assumes machine output quality is constant, and it is not. A clean segment and a badly mangled one would pay us the same, which is a pricing model that penalises exactly the work you are sending us.

S08

Ask about the exception tier

Tell us what your pipeline flags, in which languages, and at what volume. We will tell you how we would staff it.

Ask about the exception tier