Full automation tasks, from choosing the right translators to final delivery – how AI helps

By Felix Donoso

Felix Donoso is a computer engineer with a master’s degree in strategic and innovation management. With almost 20 years of experience in the sector, he has gone through different roles within the translation and localization industry. At Straker, he is focused on finding and implementing the most optimal and effective solutions for the problems our clients are facing.

The translation and localization industry, like other sectors, has suffered due to the constant pressure to optimize the production costs. The challenge is the famous OTOCOQ (“On Time, On Cost, On Quality”) This challenge within the industry has created the need to translate more and more content within the shortest time possible, with the means to increase productivity.

Within the translation industry, a large percentage of this reduction in time has been applied to the supply chain, with translators being the most affected. Partly because it was the quickest way to maintain acceptable margins as this way, it is not required to apply any technological or process changes.

However, many translation businesses have opted for another way. That is, to automate the management and operational processes as much as possible, where the human touch does not give any added value to the processes. At Straker Translations, we have addressed this problem from a different point of view.

Extraction / Automatic return of the contents to be translated:

Through the use of our custom-built API, we offer alternatives so that our clients can automatically transfer the original content from their content managers to RAY (our Ai powered technology platform) and retrieve the translated material. Moreover, for a series of CMS such as WordPress, Adobe Experience Manager, Adobe Magento. Your custom-built API delivers your content fast – to get your translations into the translation process quickly and easily.

Automation of the internal translation process:

These would be processes such as the creation of bilingual work files, the associated translation memories, etc. Also, at Straker, we have our Ai powered RAY technology platform. The automation of these processes is even more straightforward, allowing fully automated management since the original content arrives until its translation. For instance, the automatic transmission of content in each of the phases of the translation process, including allowing access to the final client to validate the translated content and its subsequent delivery.

Automation of administrative processes:

At Straker, we have automated our administrative processes through our Customer Dashboard – DELTARAY. Our customer dashboard is where you can order a translation, track progress, assign users, give access to your validators, get reports and make payments – alongside a comprehensive range of other time-saving functions.

With all these measures plus the technological advance produced in the CAT tools and in the MT engines, we might come to think that the limit of optimization of the productive process has come to an end, which is not true. There is still enough space, thanks to the use of Ai.

The first process where Ai can help us is in the process of selecting new translators. Usually, the classic process consists of validating a CV and classifying translators by the different topics in which they are said to have experience. Depending on these topics and language combinations, jobs are assigned. But is it really objective? How do we know if a translator is really specialized? Besides the quality, is it fast enough?

At Straker, we have developed a process that, starting from a natural language recognition system, we classify the subject of the first jobs carried out by a translator objectively. Based on the quality control that is mandatory in these initial jobs, we objectively discover which are the areas where the translator really meets the expected quality. In this way, we create a “scoring” where we take into account parameters such as quality and speed in a specific subject matter.

These “scorings” are those that will allow us to assign preferred translators for different clients or subject matters so that the projects will be assigned prioritizing these translators. If in a specified timing there was no answer, we would pass to the next group of translators where, depending on their scoring, additional quality control would be launched. This control would not have been carried out if they were preferred.

Other parameters to be taken into account when assigning translators to a project are geographical availability (to obtain a faster response), speed of translation and, of course, cost. But, beware, not only the direct cost at a specific moment but also the cost depending on the expected production in the following hours/days optimizing the selection. In this way, we ensure that at every moment we can use the best possible translators thanks to this predictive model and comply with the famous OTOCOQ.

Finally, all this information is complemented with the added value of being able to collect the productivity of the teams of assigned translators, to ensure the OTOCOQ is achieved but also to ensure productivity increases and the added goal of translating more content in less time and spending the same.

Our goal is to continue improving our processes by using Ai if necessary, to continue meeting and exceeding the needs and requirements of our customers.