Articul8 has launched Bivia, a multimodal decision model built to help enterprise AI agents make structured judgments across text, images, documents, tables, charts, audio, and video. Available via API, Bivia handles repeated decisions such as tool selection, evidence verification, and task completion without relying on a generative model for every step. It supports five decision types: choice, Boolean, score, set, and span. Organizations can specialize the model with their own data, policies, standard operating procedures, and historical decisions. Bivia grounds answers in source evidence by pointing to image regions, text passages, or video timestamps and can abstain when evidence is insufficient. It offers Fast, Fast+think, and Deep modes and supports languages including Hindi, Gujarati, Tamil, and Sanskrit. In internal tests the company reported higher workflow completion rates after domain training.
Articul8 has launched Bivia, a multimodal decision model built to help enterprise AI agents make structured judgments across text, images, documents, tables, charts, audio, and video. Available via API, Bivia handles repeated decisions such as tool selection, evidence verification, and task completion without relying on a generative model for every step. It supports five decision types: choice, Boolean, score, set, and span. Organizations can specialize the model with their own data, policies, standard operating procedures, and historical decisions. Bivia grounds answers in source evidence by pointing to image regions, text passages, or video timestamps and can abstain when evidence is insufficient. It offers Fast, Fast+think, and Deep modes and supports languages including Hindi, Gujarati, Tamil, and Sanskrit. In internal tests the company reported higher workflow completion rates after domain training.