Skip to main content

Breakthrough technology for managing and purchasing translation services guaranteBreakthrough technology for managing and purchasing translation services guaranteeing transparency, savings and quality

Periodic Reporting for period 1 - AQRATE (Breakthrough technology for managing and purchasing translation services guaranteBreakthrough technology for managing and purchasing translation services guaranteeing transparency, savings and quality)

Reporting period: 2018-08-01 to 2018-11-30

For companies around Europe which require thousands of words translated per week, the reality can be complex and costly. Thousands of companies are spending over €50,000 per year on translation, with many spending over €100,000. Multiple types of translations and various pricing models (per line, per word, per hour) can make it difficult for translation buyers to evaluate or compare the cost estimates provided by language service providers (LSPs).
Aqrate provides companies and LSPs with a transparent way to do translation pricing, gives companies control of their translations and intellectual assets and allows them to compare and use multiple vendors, while giving LSPs a more standardized, low-risk way of pricing their work, as well as access to a larger client market. Our market tests have shown that our service can deliver up to 30% (or more) savings to translation buyers on their annual translation costs. Aqrate’s unique value proposition goes far beyond mere cost savings, however, to encompasses the entire translation process and deliver value and benefits to all stakeholders: translation providers and buyers.
Our approach is important for society, because we seek to integrate machine translation (MT) and artificial intelligence (AI) tools for repetitive tasks (what software is ideally suited for), while preserving the human element for the linguistic, technical and creative aspects of translation work (what humans are ideally suited for). As AI and MT continue to improve, this approach will ensure that there will always be a place for people in the translation process. In addition, we estimate that we will create 130 high-value jobs in software development, sales, language services and linguistic studies, thus becoming an important employer for trained linguists (i.e. walking the walk, as far as human resources are concerned).
Our objective is to provide Agrate’s software-as-a-service (SaaS) to serve as bridge between translation buyers and LSPs. This will require building up our human linguistic resources, while at the same time improving our software algorithms to perform front- and back-end analysis of texts.
During Phase 1 Feasibility Study we have performed an extensive market analysis, including estimates potential market size and analysis of our competitors’ offerings. This has allowed us to develop our commercialisation strategy, including revenue and pricing models. The financial projections prepared demonstrate that we have a robust business model and have led us to the conclusion that we should pursue this project via a Phase 2 SME Instrument submission, while risk analysis has revealed that the risks are manageable for the execution of our plan.
Technically, we have refined the definition of our minimum viable product, in line with feedback we have had from potential customers and beta testers. A development plan for arriving to that specification has been drafted and will be further developed during our Phase 2 project.
A patent search has revealed no obstacles to our freedom to operate and exploit our innovation in Aqrate. We have also examined all the various legal aspects associated with our project ambitions, including trademarking our name, copyright issues relating to translated content and our compliance with GDPR regulations.
In our beta test cases during Phase 1, we were able to deliver more than 50% savings to a client on their annual translation costs. This is far above our target of up to 30%.
We initially began writing our software algorithms to perform source and target text analysis with the objective of coming up with accurate word counts for translation. In the course of performing this work, we discovered that a key issue was “dirty” source text, i.e. source text with inconsistencies and errors that created stumbling blocks and decision points for translators. The implication of this is huge, when one considers that a source text might be translated in dozens, if not hundreds of languages, with each translator being confronted with the same inconsistencies, which means time and money. We realized that we could achieve a significant multiplier effect by improving our algorithms to recognize and flag such source text errors, which our linguistic experts can then address via their expertise, or, if necessary, by resolving it with the client. Translation buyers get back an improved source text and LSPs get a clean text to disseminate to translators for each target language.
Many translators and linguists worry that machine translation (MT) and artificial intelligence (AI) will eventually eliminate their jobs and put them out of business. Our view is more positive. We do see AI and MT being able to handle more and more of the rote, repetitive type of linguistic work, but we see this technology as something to integrate into translation work to relieve translators of the less challenging, less intellectually satisfying work. Our view is that, while the nature of the work may change, demand for translation will only increase with advancing globalization (market studies bear this out) and that there will always be a place in the translation process for humans. Our approach to our software and our business is based on this concept.