Auto Adapt (Orchestrator)

Content is machine translated from English by Phrase Language AI.

Auto Adapt is a Phrase-developed generative AI capability.

Auto Adapt uses Generative AI to improve and transform a translated document. It restores fluency and consistency, improves grammar and ensures that brand terminology is used accurately. Customers can transform target text to meet their brand voice and audience requirements by specifying the formality level to apply to a document, and by inputting content requirements.

Powered by an interaction with multiple of OpenAI’s latest models, Auto Adapt retrieves source and target text, the project term base, formality requirements and content requirements, and transforms content at speed. When a term base is assigned to the project, Auto Adapt respects preferred and forbidden terms defined in the term base to guide the generated output. Other term base information is not currently used by Auto Adapt.

Auto Adapt is available as an action in Orchestrator and is configurable via the Orchestrator workflow editor. The Auto Adapt with Human Review workflow template provides an example of one possible use case.

The Auto Adapt action consumes AI Units (AIUs). Customers do not need their own OpenAI credentials to start using it.

To ensure timely processing, the recommended input limit is 5,000 words.

Due to very large file sizes, .IDML (Adobe InDesign) files are not supported.

Use cases

Creative adaptation and generation of content

Auto Adapt transforms existing content to produce new content. This makes it useful for generating variants of text, such as for new audiences.
- Transform product copy to suit a new audience segment.
- Change product descriptions to appeal to buyers in a new market.
- Make subtitles and in-game text consistent in gaming and media content.
- Create targeted SEO and marketing copy for each locale and segment.
- Tailor training materials to the reading level of audience.
- Refine internal communications to meet the needs of multinational teams.
- Use inclusive and age-appropriate language for online copy.
- Correct use of domain terminology for regulated industries.
- Proofread and fix website content.
- Improve source content for localization.

Since Auto Adapt transforms content creatively, the target content processed by Auto Adapt departs from being purely a translation of the source content. While Phrase QPS captures some changes, it may not fully assess the effectiveness and quality of adapted content.

Exclude locked segments

When configuring Auto Adapt in an Orchestrator workflow, users can control whether locked segments are included in AI processing.

In the Parameters section of the Auto Adapt action, set Include locked segments as AI context to:

Yes (default)

Locked segments are included as context and consume AI Units (AIUs). Including locked segments provides more context and can improve output quality.
No

Locked segments are excluded and do not consume AIUs. Excluding locked segments may lower output quality.

This option is useful for jobs with a high proportion of locked segments that do not need to be considered during optimization.

Existing workflows must be republished for the change to take effect.

Language support

Although Auto Adapt may support many language combinations, we have determined a set of recommended language pairs selected to adhere to Phrase’s high-quality standards:

English
French
Italian
Spanish
Portuguese
German
Dutch
Swedish
Danish
Norwegian Bokmål
Norwegian Nynorsk
Finnish
Hungarian
Estonian
Czech
Russian
Polish
Bulgarian
Greek
Latvian
Chinese Simplified
Chinese Traditional
Japanese
Korean
Thai
Vietnamese
Indonesian
Arabic
Turkish
Hindi

Content Requirements for Auto Adapt

Content requirements are one of several configurable parameters that customize Auto Adapt.

Auto Adapt takes the content requirements provided and uses them in its interaction with OpenAI’s generative AI models. Specifically, the exact contents of the field are interpolated into the mechanisms used to interact with the different models, rendering the input text as a list; each sentence of the content requirements is an item in the list.

When setting up content requirements for the Auto Adapt action, the maximum length is 4,000 characters.

For best results, Auto Adapt looks for three key pieces of information:

Brand voice
Intent or purpose of the content to be processed
Target audience

Best practices

Be concise and specific

Use as few words as needed to communicate the content requirement. Avoid vague instructions like make it sound better, as they do not translate well into model behaviors. Using one sentence per requirement is recommended.
Avoid references to Translation Memories (TMs) or other external resources

Auto Adapt does not integrate with TMs. Instead, provide explicit behavioral examples where needed.
Avoid redefining model roles

Auto Adapt is already optimized to act as an expert reviewer of [language] to [language]. Requesting that the model take on a different role is not recommended.
Avoid requesting built-in capabilities

Auto Adapt optimizes for fluency, terminology, consistency, and tag handling natively. There is no need to duplicate these requests.

Example Output Scenarios

Neutral input text:

Our new product has several innovative features designed to improve user experience. It is easy to set up and can significantly enhance productivity in everyday tasks.

Scenario: Clarity is key
- Content requirements:
  
  Focus on clear, concise, and direct language. Avoid jargon and overly complex sentences. Prioritize readability and simplicity, using plain language wherever possible. Maintain a logical flow and ensure that the message is easily understandable for a broad audience.
- Auto Adapt output:
  
  Our new product offers innovative features that make it easier to use. Setting it up is simple, and it helps boost productivity in everyday tasks.
Scenario: Technical precision for technical product documentation
- Content requirements:
  
  Adopt a formal and precise tone, suitable for technical or scientific documentation. Prioritize accuracy, ensuring that terminology and explanations are consistent with industry standards. Avoid ambiguity by using specific and well-defined terms. Use clear structure to present information logically and thoroughly.
- Auto Adapt output:
  
  The new product incorporates advanced features aimed at optimizing user experience. Its straightforward setup process and functionality enhancements significantly improve task efficiency.
Scenario: The friendly yet professional touch
- Content requirements:
  
  Use a warm and professional tone that builds trust and rapport. Balance friendliness with clarity, ensuring that the message remains informative and approachable. Avoid overly casual language while maintaining a personable style. Keep sentences structured yet inviting, emphasizing a customer-centric perspective.
- Auto Adapt output:
  
  Discover our new product, designed to make your life easier with innovative features. Simple to set up and use, it helps you get more done every day.
Scenario: Informal and witty
- Content requirements:
  
  Adopt a conversational and light-hearted tone, incorporating humor and playful expressions when appropriate. Use casual, engaging language to make the content relatable and enjoyable. Prioritize creativity and originality while keeping the message clear and contextually appropriate for social media or blogs.
- Auto Adapt output:
  
  Say hello to our awesome new gadget! It’s packed with cool features that make life way easier—setup’s a breeze, and it seriously ups your productivity game.
Scenario: For a young audience
- Content requirements:
  
  Use a dynamic and lively tone that resonates with younger audiences (teens and young adults). Incorporate contemporary slang and cultural references where appropriate, while maintaining clarity. Keep the style informal, upbeat, and engaging. Prioritize relatability while avoiding condescension or overly formal language.
- Auto Adapt output:
  
  Check out our new product—it’s packed with awesome features to make life easier. Super easy to set up, and it’ll totally level up your day-to-day hustle.

Template Library

Auto Adapt’s prompt architecture interprets Content Requirements as a controlled style instruction, so the patterns that work best are:

Imperative form (e.g. “Refine the translation to ensure…”).
Clear audience and intent markers (“for HR managers”, “marketing copy”).
Consistent single focus per clause, avoiding meta-instructions like “first do X then Y.”
Concrete quality signals (fluency, clarity, tone, terminology).

Use Case Examples

Use Case	Source text	Pre-translation (Automated translation)	Content Requirements	Result
character limitation	In this particular moment, I find myself compelled, with an almost uncontrollable urge, to explain in a somewhat excessively detailed and unnecessarily elaborate manner, that this sentence has been purposefully made overly verbose.	En este momento en particular, me veo obligado, con un impulso casi incontrolable, a explicar de una manera un tanto excesivamente detallada e innecesariamente elaborada, que esta oración se ha hecho deliberadamente demasiado verbosa.	It is absolutely vital that the translations for each segment do not exceed the maximum character limitation, which is 40. If the translations in any given segment have more than 40 characters, you must edit the translation so that it contains 40 characters (at the most) while also keeping the same general meaning.	En este momento debo explicar que esta frase es demasiado larga.
time zone conversion	We are available from 8 AM EST to 11 PM EST	Estamos disponibles desde las 8 a. m. EST hasta las 11 p.m. EST.	Adjust machine translations to convert time and date information from EST to the correct local timezone for each target language, ensuring accurate localization. Instructions: 1. Identify time and date: Detect all time, date and timezone references in the source text. 2. Convert to target timezone: Apply the following rules per locale: - es_MX: convert EST to PST. PST is 3 hours behind EST. Format: PST.	Estamos disponibles desde las 5 a. m. PST hasta las 8 p. m. PST.
regional adaptation	As part of these reforms, citizens will have to pay extra postage with a special designated color for any delivery made outside a 500 kilometer range.	As part of these reforms, citizens will have to pay extra postage with a special designated color for any delivery made outside a 500 kilometer range.	Fix any errors in the document, ensure consistency, fix tag and formatting issues. Please implement British English spelling. This includes, but is not limited to changing "z" to "s (example: nationalize --> nationalise), adding "u" in certain words (example: color -> colour). This list is not exhaustive, so please make any other changes necessary.	As part of these reforms, citizens will have to pay extra postage with a special designated colour for any delivery made outside a 500 kilometre range.

Auto Adapt in the CAT web editor

When opening a job that has been processed by Auto Adapt, the CAT web editor displays the Job auto adapted notification toast.

Auto Adapt post-editing can result in the following output:

Edited target

The segment text was modified by Auto Adapt. This is indicated by the Auto Adapt indication in the score/origin column of the CAT pane and the Auto adapted tooltip.