AI translation service levels, with a native linguist on every job
Not all of your content carries the same risk, so it should not all get the same human depth. Our three AI translation service levels set that depth, and every job that uses AI has a native linguist on it. AI agents translate against your glossary, translation memory (TM), and style rules. What changes from level to level is how much of the text the linguist works on: at AI + Check, every segment our agents flag plus a random sample; at AI + Edit, every segment, edited by a native post-editor; at AI + Edit + Review, every segment, edited and then reviewed by a second linguist.
You choose a level for each content type, and your quote names it. Archived help articles, product UI, and campaign pages each run at the depth their risk calls for, so linguist time goes where your readers will notice it.
- AI + Check Accurate, consistent and understandable at scale 1234567
- AI + EditDefault Reads as if it was written in the target language 1234567
- AI + Edit + Review For high-visibility or high-risk text: edited, then reviewed by a second linguist 1234567
Three levels of human depth, chosen per content type
The table shows what runs at each level, what you receive, and which content each level fits. A segment is one sentence or one string.
| Row | AI + Check | AI + Edit (default) | AI + Edit + Review |
|---|---|---|---|
| Outcome | Accurate, consistent, and easy to understand | Reads as if it were written in the target language | For high-visibility or high-risk text: edited, then reviewed by a second linguist |
| Human work | A native linguist checks every flagged segment plus a random sample, fixes errors, and settles every flag | A native post-editor edits every segment. A reviewer, a second linguist, then checks a sample of the edited text | A native post-editor edits every segment. A second linguist then reviews every segment, not a sample, and does a final read of the delivered file |
| Agent checks | Native-reader agent, plus an accuracy agent that flags meaning risks | Native-reader agent, plus quality assurance (QA) agents that check accuracy, grammar, style, and terminology, each blind to the others | Everything at AI + Edit, plus independent term research, a red team, a debate between advocate and devil’s-advocate agents, and a regression check on changed text |
| Cycle steps | Steps 1–4 and 7 | Steps 1–5 and 7 | All seven steps |
| Who decides findings | The checking linguist | The post-editor, or the reviewer for text the post-editor changed | The reviewer, after a judge agent recommends |
| What you receive | Files in your format, the quality-gate result, an issue log, and your updated TM | The same, plus glossary changes and a query log, and sampled quality reports on the schedule agreed for your program | Everything at AI + Edit, with a full quality report on the whole file with every delivery |
| Content it fits | Internal knowledge, archives, user reviews, long-tail catalog attributes, low-traffic help articles | Help-center articles, support macros, UI strings, website pages, catalog titles and descriptions | Campaigns, brand pages, and regulated and high-risk content |
| Never used for | Regulated, legal, safety, or financial text; UI strings; checkout and payment text; headlines and brand copy | Regulated and legal text, safety-critical text, financial disclosures | — |
Every job runs steps 1–4 and 7. Steps 5 and 6 run at the higher service levels, as named in your quote.
Sample sizes are agreed with you in the quote, and every issue log and quality report states how its sample was drawn and how large it was.
At every level, your instructions and glossary outrank the linguist, and the linguist outranks any agent.
If you buy machine translation post-editing (MTPE) today, compare it with AI + Edit. Beyond a full edit by a native post-editor, it adds AI calibrated on your glossary, TM, and style rules; terms sourced from your own content; agent checks before and after the post-editor; automatic checks for the rules that can be tested reliably; a quality gate on every delivered file; the gate result, an issue log, glossary changes, and a query log with every delivery; and sampled quality reports on an agreed schedule. AI + Edit + Review adds a second linguist’s review of every segment on top. See what our AI post-editing adds to MTPE.
AI + Check
Accurate, consistent, and easy to understand.
AI + Check is AI translation with a linguist check. Agents translate against your glossary, TM, and rules. A native-reader agent then flags any text that reads like a translation, and an accuracy agent compares source and target to flag meaning risks. A native linguist checks every flagged segment plus a random sample, fixes the errors, and settles every flag. The linguist’s time goes where the risk is, and unflagged segments are delivered as AI output that has passed the automatic checks and the quality gate.
This level is not post-editing. It suits content that readers need to understand quickly and correctly, where nobody is judging the prose: internal knowledge, archives, user reviews, and long-tail catalog attributes. It never runs on regulated text, UI strings, checkout and payment text, or brand copy.
If your current vendor has a linguist read and lightly edit every segment, compare that with AI + Edit, which goes further: a native post-editor edits every segment, with agent checks before and after the edit.
You receive your files, the quality-gate result, and an issue log of what the linguist checked and fixed.
AI + Edit
Reads as if it were written in the target language.
AI + Edit is our default level. After AI translation, a native post-editor edits every segment. This is full AI post-editing (AIPE): if you buy machine translation post-editing (MTPE) today, this is the level to compare it with. QA agents then check accuracy, grammar, style, and terminology, each blind to the others’ results, and a reviewer runs a sampled quality check. The linguist decides each finding directly.
Use it for content your customers read every day: help-center and knowledge-base articles, support macros, UI strings, website pages, and catalog titles and descriptions.
You receive your files, the quality-gate result, an issue log, glossary changes, and a query log with every delivery, plus sampled quality reports on the schedule agreed for your program. One report can cover several small releases.
AI + Edit + Review
For high-visibility or high-risk text: edited, then reviewed by a second linguist.
A native post-editor edits every segment. A second linguist then reviews every segment and does a final read of the delivered file. The agents do more at this level too. Separate agents research terms independently, and a red team of agents tries to make the file fail and must back every claim with evidence. Advocate and devil’s-advocate agents argue each finding, a judge agent recommends, and the reviewer decides and records the reason. A script applies the agreed changes, and a regression check re-reads everything that changed.
Use it for campaigns, brand pages, and regulated and high-risk content. Your quote states the specialism the job needs. If you buy post-editing followed by an independent review today, this is the level to compare it with.
You receive your files, the quality-gate result, an issue log, glossary changes, a query log, and a full quality report on the whole file, with every delivery.
Which level fits which content
Use this as a starting point. You make the call for each content type, and the quote records it.
Start from the impact of an error:
| If an error would | Typical content | Level |
|---|---|---|
| Slow a reader down, in content nobody judges for style | Internal knowledge, archives, user reviews, long-tail catalog attributes | AI + Check |
| Be noticed by customers, or weaken trust in your product | Help-center articles, UI strings, website pages, catalog titles and descriptions | AI + Edit |
| Cause harm, legal or financial exposure, or public damage to your brand | Campaigns, brand pages, and regulated and high-risk content | AI + Edit + Review, or no AI |
Regulated and high-risk content never runs at AI + Check. Then find your content type in the table below.
| Content type | Usual level | Note |
|---|---|---|
| Internal knowledge and archives | AI + Check | |
| User reviews and other user-generated content | AI + Check | |
| Archived or low-traffic help articles | AI + Check or AI + Edit | Your choice, per article tier |
| Long-tail catalog attributes | AI + Check | Never attributes that carry safety or compliance data |
| Help-center and knowledge-base articles, support macros | AI + Edit | |
| Website pages | AI + Edit | |
| Catalog titles and descriptions | AI + Edit | |
| UI strings, checkout and payment text | AI + Edit or AI + Edit + Review | Never AI + Check |
| Campaigns and brand pages | AI + Edit + Review | |
| Regulated and high-risk content | AI + Edit + Review, or no AI | Never AI + Check |
Anything with legal effect, such as terms or warranties, never runs at AI + Check. For regulated and high-risk content, your quote names the level or the no-AI route, and the subject specialism the job needs.
What each level delivers
Every delivery includes the quality-gate result and an issue log. At AI + Edit, sampled quality reports follow the schedule agreed for your program and can cover several small releases; at AI + Edit + Review, each delivery gets a full report.
- Files in your format, with tags and placeholders preserved
- The quality-gate result for every delivered file
- An issue log of what was found and fixed
- Sampled quality reports at AI + Edit, on the schedule agreed for your program
- A full quality report with every delivery at AI + Edit + Review, covering the whole file
- Glossary changes and a query log at AI + Edit and AI + Edit + Review
- Your TM, updated, with a Translation Memory eXchange (TMX) file on request
- Under a monthly contract, a monthly summary of quality and of deliveries against the agreed windows
The quality report classifies issues as critical, major, or minor, across accuracy, fluency, terminology, style, locale, tags and format, and client requirements. It carries the label “Assessment by Locstars.” Scores appear only in your own reports, never on this site.
An automated quality gate runs on every delivered file. If it fails, the file does not ship. If the gate flags something that is not an error, we fix the gate’s rule and record the case in your delivery note. We never edit your file just to make it pass.
Prefer no AI
Some content should not go through AI, and some company policies rule it out. Tell us which content, and it is translated by native linguists without machine translation or generative AI, using your translation memory and glossary. Every language we offer has a no-AI route.
Regulated and high-risk content runs at AI + Edit + Review or on the no-AI route, never at AI + Check. You can mix routes in one program: AI + Edit for your help center, for example, and no AI for your contract templates.
Questions about service levels
Can we use different levels for different content?
Yes. You choose a level for each content type, and the quote names it. One program might run archived help articles at AI + Check, current articles at AI + Edit, and campaign pages at AI + Edit + Review.
Is AI + Check a form of post-editing?
No. At AI + Check, a native linguist checks every segment the agents flag, plus a random sample, and fixes what is wrong. If you want a native post-editor to edit every segment, choose AI + Edit.
Which level is the default?
AI + Edit. A native post-editor edits every segment, QA agents check the result, and a reviewer runs a sampled quality check. It is the usual level for help-center, UI, and website content.
Does every level include a quality report?
Not at AI + Check. Every delivery includes the quality-gate result and an issue log. At AI + Edit, sampled quality reports follow the schedule agreed for your program and can cover several small releases; at AI + Edit + Review, each delivery gets a full report. AI + Check deliveries do not include a quality report.
Who has the final say on a change?
Your instructions and glossary do. Then comes the deciding linguist, and then, at AI + Edit + Review only, the judge agent’s recommendation.
Can we get same-day delivery?
Same-day delivery is possible for suitable volumes. We confirm timing in the quote.