Large Scale Content Automation with generative AI

Large Scale Content Automation (LSCA) refers to the use of generative AI to automate content creation in an international commercial environment. By combining data-based and dynamic text generation in the content automation platform textengine.io, LSCA offers a scalable method to produce web, product and SEO texts in all common European and selected Asian languages. LSCA is based on initially set up, data-driven text models which provide a rule-based text framework to ensure the output is content-rich, grammatically correct, and legally compliant.

LSCA targets the text source by using human feedback to improve the relevant text models thematically. In cases like ChatGPT, the text automation occurs at the individual employee level, without strategically organising content production or content workflows. A holistic business process automation through LSCA goes much further, as automated text generation is designed for all desired use cases and text types.

LSCA is designed for digital commerce, media and beyond, targeting marketing, content, and SEO teams. Both B2C and B2B businesses can use it to implement scalable content automation across all available data, accessing predefined text formats and processing custom prompts. For a consistent corporate language, content, wording and tone can be auto aligned to desired brand guidelines. Human involvement remains a crucial aspect, with users retaining control over the text output (Human-in-the-Loop).

Customer benefits:

• Quick time-to-market by using structured and unstructured data
• Unique content for SEO, cross-channel content production and rewrites
• Consideration of customer buying intent automatically
• Support for a range of text genres (product descriptions/detail pages, category/catalogue pages, target/POI content, FAQs, HowTos and SEO texts)
• Legally compliant and error-free content generation when using generative AI
• Effective and resource-saving content workflows for multilingual language projects

Large Scale Content Automation allows for a more holistic approach to business process automation than systems such as ChatGPT, facilitating automated text generation for all use cases in European and Asian languages across the value chain. It ensures consistently high content quality even when producing 10,000 or 100,000 of web, product and SEO texts, while also considering SEO and compliance standards, and allows for the creation of personalised and conversion-efficient content. The inclusion of data-driven text models allows for virtually limitless scaling.

The unique solution provided by LSCA uses generative AI to provide high-quality texts on a large scale in various languages. It ensures that the content is error-free and allows companies the flexibility to adjust according to their needs. The use of generative AI with machine learning allows for the incorporation of global knowledge into text generation processes.

Compared to providers of large language models such as OpenAI or Google, which pursue a 1:1 approach with a human in the quality assurance and fact-checking loop, Large Scale Content Automation opens up more economical possibilities through a 1:N approach - one-time setup for x times application. This leads to faster and easier text generation, making the process more efficient and cost-effective.

Overall, LSCA significantly expands the strategic options for content automation in companies targeting different international markets by producing a variety of text genres easily and effectively. This is an innovation in the field of business process automation and not at the level of individual employees or smaller teams. Rather, a central solution is enough to address a wide variety of use cases in the field of text automation by generative AI, in order to launch a high-quality and easily scalable content production.

Hybrid NLG 02/23: Hybrid NLG is an integration between our data-driven language model and end-to-end text generation based on GPT-models. With the integration of GPT-models into the textengine.io, customers can generate example texts based on a few inputs. The setup and use of text models can be significantly accelerated, as the time-consuming researching and formulating of information is no longer necessary. Variant Suggestions 03/23: textengine.io offers automatically generated text variant suggestions for different languages. Companies get access to an AI-supported assistant that offers unique alternatives for each formulation without changing the content of the statement. Rewrite 05/23: With the rewrite functionality, contents can be automatically rewritten and varied in different languages. Rewrite is designed to transform existing content, enrich it with data and individual information, trim it to the desired tone and length, and optimise it for search engines. Textmodel Creator 07/23: The Textmodel Creator offers a clever AI assistant that helps to create fully functional text models within minutes. Customers benefit from automated creation processes that dynamically provide all necessary structural elements (Messages, Template, Text-Variants) - and which can subsequently be fine-tuned manually. Message Composer 10/23: The feature enables customers to efficiently create automated texts without having to write themselves. This is possible thanks to the ability to convert a sample sentence for a "saved prompt" into a fixed template. By converting dynamic templates into static ones, near-perfect text compliance is achieved. Buying Intent Assistant 11/23: With this feature, companies can create automated texts that are specifically tailored to the purchasing intentions of their customers. The Buying Intent Assistant creates automated text suggestions that consider the most common Google search queries and keywords on a specific topic or product category.

Lyreco, an international provider of workplace solutions, has partnered with us since 2019 to automate the creation of product descriptions in the Swiss market in four languages. By 2023, their goal was to implement this automated generation of product descriptions across 20 languages worldwide, including selected Asian languages. The objective was to streamline business processes, save costs, increase efficiency and reduce time-to-market when introducing existing product groups.

Retresco won a public tender and collaborated successfully to automate product descriptions in over 20 countries and languages in 9 months. The automated product descriptions were launched in places like the UK, Western and Eastern European countries, and several Asian markets including Hong Kong, South Korea, Thailand, Malaysia, and Singapore. The automation system uses one-time setup text models that consider technical product attributes and ensure accurate textual variance. These text models can be used unlimitedly and adapted to changing product and market requirements.

The text models for all 20+ target languages were set up in textengine.io in a short time period. This collaboration has reportedly led to a 96% reduction in content creation time compared to manual processes. In the future, Lyreco is hoping for a 30% boost in country-specific conversions, like the success seen in the Swiss market. 10,000 detailed, search engine optimised product texts were produced in a matter of mere minutes.

“The quick international rollout of our product information is something we are very pleased about. This would not have been possible without Retresco‘s expertise in business process automation and regular support. We centrally designed and initiated the content workflows, so that our 20+ country organisations can work autonomously and productively with them. As a result, Lyreco saves resources and effort and further increases its efficiency.” Arnaud Delepouve, Group Content Manager at Lyreco

https://www.retresco.de/