Hybrid NLG

The leading provider of AI-based content automation Retresco enables the simple and automated generation of web, SEO, and product texts with the help of Artificial Intelligence ("Natural Language Generation"). The content automation platform textengine.io is used by many renowned companies in the area of digital commerce such as MediaMarktSaturn, Bauhaus, Toom Baumarkt, Hellweg, ImmoScout24, Targobank and the Spiegel publishing house.

On 28 September 2022, Retresco launched the innovation "Hybrid NLG" for simple content creation. This allows the data-based approach to text automation based on parameterised templates to be combined with the advantages of large language models such as GPT-3. This technology uses Deep Learning to capture freely available internet content and train itself with the knowledge required for text generation. Retresco now integrates this world knowledge with the possibilities of its content production, which is designed to scale.

With Hybrid NLG, companies manage demanding content projects much more easily and effectively, and the time-to-market is significantly reduced. This applies not least to multilingual projects, as all common European languages are supported. Through the seamless integration of large language models such as GPT-3 into textengine.io, companies benefit from the latest possibilities in AI-powered content automation combined with text variance for convincing copies as well as high SEO relevance. However, the Retresco innovation is not only designed for a fast implementation of language projects, but also enables effective creation processes when it comes to grammatically and lexically flawless and search engine optimised texts in different eCommerce and marketing channels.

The most important customer benefits are:

• New options for faster time-to-market
• More options for different text types
• Integrated world knowledge and automatic style
• Unique content for SEO and omni-channel use
• Faster and easier multilingual content creation
• Legally compliant and error-free use of Natural Language Generation

Overall, Hybrid NLG offers new strategic options in text automation. In the future, companies will only need one solution in Natural Language Generation for various use cases involving high-quality and easily scalable content production. Natural Language Generation is thus moving to the centre of text-based business processes – and thus to a strategic tool along the entire digital value chain.

With the launch of Hybrid NLG, we combine automatically generated text suggestions via GPT-3 with the data-based text models and possibilities of the content automation platform textengine.io. The central advantage here is that the text models essential for automated content creation can be set up much faster. At the same time, Hybrid NLG supports a wide variety of text types: Product descriptions, category/catalogue pages, destination/POI content, product comparisons and FAQ content. At the same time, content can be laid out even more granularly, textually varied and enriched with world knowledge freely available on the internet.

Hybrid NLG supports companies in practice by automatically generating sample texts based on a few inputs. This significantly speeds up the set-up and use of text models, as it eliminates the time-consuming research and formulation of information. In internal tests, Retresco was able to determine a time saving of up to 50 % during the onboarding phase in the first 3 months when setting up and using text models. In addition, data-based content production becomes even more user-friendly and customisable.

AI-powered Automation of Product Descriptions with MediaMarktSaturn

The MediaMarkt and Saturn online shops consistently offer more than 300,000 products. And while these online shops provide customers with an enormous selection of different products, for the online team the extensive product range means one thing above all: copywriting. After all, each product requires a comprehensive description – without it, customers would struggle to make informed purchasing decisions.

The product copy is based on information provided by manufacturers. In the conventional copywriting process, however, many manufacturers primarily provided technical specifications, not marketing copy, which led to inconsistencies in the product descriptions on the website. Additionally, MediaMarktSaturn was confronted with a further challenge faced by all large online retailers: The more products a shop contains, the more time-consuming it is to write unique product descriptions.

Since 2019, the company’s online shops have relied on automatic text generation using the software textengine.io from Retresco. textengine.io is a natural language generation (NLG) application. NLG is a branch of artificial intelligence. Data forms the foundation of the NLG application; templates and conditions also need to be defined in advance to produce text in natural language. If these prerequisites are met, the software generates unique content using data-based product characteristics such as colour, size, material, or other individually defined properties.

By using textengine.io, MediaMarktSaturn now generates more than 300,000 product descriptions. Since its launch, textenengine.io has been used intensively for the online shops of MediaMarkt and Saturn. In addition to automatic product descriptions, all services offered online around the articles are to be described with automatically generated texts in the future. Overall, the content team benefits from a low use of resources with maximum output.

The content automation platform textengine.io is regularly used by more than 100 Retresco customers from a wide range of industries, both nationally and internationally. The spectrum of use ranges from e-commerce providers and retailers to industrial companies and financial service providers to media houses and comparable publishers. In digital commerce, for example, automated text generation helps companies to cover their enormous demand for content in a wide variety of formats, such as product descriptions, advertising copy and SEO texts, in an automated way and to scale them flexibly. In all application scenarios, structured data is converted into high-quality texts.

Hybrid NLG has met with a great and positive response from users a wide range of companies since its launch at the end of September 2023. Hybrid NLG is already regularly used by Retresco customers such as MediaMarktSaturn, Jungheinrich, BBV-Domke and ProSales24.

Hybrid NLG is seamlessly integrated into textengine.io and creates automatically text suggestions for a wide variety of text types – for example, to launch tens or hundreds of thousands of FAQ pages, category and destination pages or product descriptions. The combination of data-based content automation based on text models and end-to-end generation with a large language model such as GPT-3 gives companies even more speed in the implementation of content projects as well as more textual flexibility in the automated creation of texts that are convincing, distinguishable and that are also found in the cross-channel.

https://www.retresco.de