Draft, review, export

AI-assisted translation workflow — a model proposes, a person releases.

An AI-assisted translation workflow uses a model to propose a draft and keeps a person in charge of what is approved. The sequence is context, a first draft, human review, a check someone can see, and an export. Aquilla runs that sequence in one workspace. It does not pull files from a CMS, assign a vendor, or publish on its own.

The direct answer

What is an AI-assisted translation workflow?

It is the order of work when a model helps and a person still owns the result. You set who the translation is for and what must be reviewed. The model drafts. Someone who knows the subject corrects meaning, not just fluency. You export what was approved. Enterprise tools add intake from other systems and a push back to those systems. Aquilla stops at the workspace: no connector, no vendor desk, no unattended publish.

Smartling, Phrase, Wordbee, and Trados all describe this shape. AI or machine translation produces a first pass. Translation memory and terminology steer it. A human reviews. Quality checks run. Someone approves, and a file or a connector delivers the result. The products differ in how much of that path is automatic.

Aquilla is the part where drafting and review happen together. Agents propose from project direction and from decisions you have already authorized. Experts keep authority over meaning and release. That is not a hub of machine-translation engines, and it is not a visual builder that routes content through a CMS.

What the search is actually choosing

An AI-assisted workflow is a job list — not “turn on the model.”

Searchers mix a first draft, post-editing, automated QA, and delivery back into a CMS. Aquilla covers the draft and the review, and says no to the rest.

JobAquillaTypical result for this search
Set the audience, scope, and what must be reviewedYes — the owner sets direction and the approval boundary.Yes — a brief, a style guide, or project instructions
Pull source files from a CMS, repo, or design toolNo — no CMS, Git, or Figma connector.Often yes — Phrase, Smartling, Smartcat, and similar platforms
Apply approved wording to the next draftYes — Living Memory learns from decisions you authorize.Yes — translation memory and a term base
Choose among many machine-translation enginesNo — Aquilla is not an engine hub.Yes — Smartling, Phrase, and Trados-style AI connectors
A model proposes the draftYes — agents propose. People approve.Yes — machine translation or an LLM first pass
Human post-editing as the whole jobNo — experts decide consequential points, not every segment by rote.Often yes — that is machine-translation post-editing
Assign a vendor and report cost per wordNo — no vendor marketplace and no cost desk.Typical of a translation management system
A check a reviewer can inspectYes — support evidence and consistency a person reviews.Automated QA plus a human pass
Push the finished file back to a CMSNo — export from the workspace. No connector push.Often yes
Publish with no one approvingNo — people release.Some platforms offer unattended delivery. Aquilla does not.

Run the sequence without inventing a platform you do not have

Context before the draft

The owner names the audience, the constraints, and which decisions need a person. Agents then draft from that direction and from wording the team has already approved. A model with no project context is not this workflow.

Review is a decision, not a cleanup queue

Machine-translation post-editing asks a linguist to repair segments one by one. Aquilla asks an expert to decide the points that change meaning, keeps that decision, and can apply it to later work. The expert still validates the result.

What a TMS or CAT package still does

Intake from a CMS, a vendor pool, per-word cost, and a desktop package from Trados or memoQ are different products. Aquilla does not open those packages and does not report vendor spend. Use the translation management page for the platform question, and the professional tools page for the desktop editor.

Scripture adds stages this page does not replace

Bible translation still needs team checking, community testing, a back translation, and a consultant when the project requires them. Those stages have their own pages. This page is the general AI-assisted sequence, not a Scripture quality manual.

Which page to use

This page is the sequence. The others answer a different question.

Agentic translation operations explains how an agent works inside permissions. Expert-in-the-loop explains where a person’s judgment has leverage. The translation management page is the platform, including what Aquilla will not connect to. Bible translation workflow is the Scripture quality sequence.

If you searched for the steps from an AI draft to an approved export, stay here.

Common questions

AI-assisted translation workflow FAQ

What is an AI-assisted translation workflow?

It is a sequence: set the context, let a model propose a draft, have a person review what matters, check what you can see, and export only what someone approved. Smartling, Phrase, Wordbee, and Trados describe that sequence inside a localization platform. Aquilla runs context, draft, review, and export in one browser workspace. It does not connect to a CMS or publish on its own.

How is this different from machine translation post-editing?

Post-editing treats the model’s output as the draft and the linguist as the person who repairs it segment by segment. An AI-assisted workflow can start the same way, but Aquilla does not make post-editing the job. Experts decide consequential points, those decisions stay in the project, and agents propose where they apply next. A person still validates before release.

Does Aquilla connect to a CMS, Git, or Figma?

No. There is no connector that pulls source files in or pushes a finished translation back to a CMS, repository, or design tool. Bring the work into the workspace and export when you are ready. If a connector is the requirement, this is the wrong tool.

Does the workflow assign vendors or price the job?

No. Aquilla does not run a vendor marketplace, does not assign linguists for you, and does not report cost per word. The people who review are the people you already trust. A translation management system that exists to route vendors is a different product.

How is this different from a translation management platform?

A translation management platform is the system around the work: where files came from, who was assigned, and often what it cost. This page is the sequence inside the work: context, draft, review, export. Aquilla’s platform page answers the first question and states the same limits. Use that page for the platform. Use this page for the steps.

Where does Bible translation checking fit?

Scripture projects add team checking, community testing, back translation, and consultant review. Those are quality stages, not a shorter enterprise workflow. Aquilla does not replace Paratext Basic Checks or translationCore. Follow the Bible translation workflow page for the stages, and the checking page for the software boundary.

Can the AI release the translation?

No. Agents propose drafts and can flag consistency questions. A person approves consequential decisions and release. Aquilla does not offer unattended publishing.

How much does an AI-assisted translation workflow cost in Aquilla?

Anyone can start on the Free plan with a weekly allowance of AI assistance. Bible-translation teams and ETEN affiliates get team-plan access at no cost — contact us to activate your organization’s workspace.

Related topics

Let the model draft. Keep the release.

Open a Free plan and review what the agent proposes. Bible-translation teams and ETEN affiliates can contact us for team-plan access at no cost. Aquilla will not connect to your CMS or publish without a person.