Shvydko
Comment: Regurgitated company announcements (churnalism) are not independent coverage. No indication of meeting Wikipedia's notability criteria at this time. Helpful Raccoon (talk) 08:56, 2 November 2026 (UTC)
Shvydko is an artificial intelligence-based service designed for text analysis and decision-making according to specified criteria. Developed by Codebridge Technology, Inc., it became publicly available on 16 October 2026.
Shvydko provides structured results in response to queries. For example, selecting an option from a predefined list, assigning a score based on a specific scale, or determining the probability of a given statement. The service operates via an HTTP API and is designed for integration with other software systems.[1][2]
How it works
Shvydko processes text based on questions and criteria provided by the user. Using this data, the model generates a response in JSON format, which can then be utilized for further automated processing.
The service supports three types of responses: Choice (selecting one option from predefined categories), Score (evaluating information on a specific scale), and Noul (determining the probability of a statement as a number between 0 and 1).[1][3]
This approach can be applied to tasks such as classifying inquiries, verifying whether text complies with established rules, or identifying cases that require further review. Shvydko does not perform subsequent actions on its own; instead, the software system receiving the model's output determines the next step.
Access to the service is managed via an API key. A web-based "Playground" environment is also available, allowing users to test the model by submitting their own queries and reviewing the results.[2][3]
Development
Shvydko was developed by Codebridge Technology, Inc., a software development company founded in 2021.
According to dev.ua, work on the model began on 26 September 2026, and public access to the API opened on 16 October.
One of the developers' objectives was to create a system capable of processing texts in both Ukrainian and Russian, specifically messages that combine elements of both languages. During development, the team used datasets collected in Ukrainian, Russian, and surzhyk.[1]
Compatibility
Shvydko was designed with compatibility for the request format of Jev, a separate text-based decision-making system developed by TypeSafe AI.
Codebridge claims that existing integrations with Jev can be migrated to Shvydko simply by updating the API address and access key.[1]
Performance evaluation
In October 2026, Codebridge released the results of its own comparative testing between Shvydko v2 and Jev 1.14.
The evaluation utilized a diagnostic sample consisting of 1,662 questions in Ukrainian. Based on the metric employed by the company, which assigns equal weight to three categories of tasks, Shvydko achieved a score of 90.50%, while Jev scored 87.57%. Representatives from Codebridge noted that the comparison was conducted using an internal diagnostic dataset employed during the development process. It did not constitute an independent benchmark and does not prove an overall superiority of Shvydko over Jev.
The evaluation also had several limitations. Specifically, it did not assess the models' operational speed, cost-efficiency, or the quality of processing English-language queries.[1][3]
References
- ↑ 1.0 1.1 1.2 1.3 1.4 Brovinska, Mariia (22 October 2026). Українська відповідь на Jev: Codebridge за 20 днів зробила API, що ухвалює рішення за текстом українською, російською та суржиком [A Ukrainian answer to Jev: in 20 days, Codebridge built an API that makes decisions based on text in Ukrainian, Russian and Surzhyk]. dev.ua (in українська). Retrieved 2 November 2026.
- ↑ 2.0 2.1 Salivio, Sofiah Nichole (27 October 2026). "Codebridge launches shvydko.ai for software decisions". IT Brief UK. Retrieved 2 November 2026.
- ↑ 3.0 3.1 3.2 "Shvydko Decision Engine". Codebridge Technology. Retrieved 2 November 2026.
This article "Shvydko" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Shvydko. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
