Enterprise AI company Articul8 has launched Bivia, a multimodal decision model designed to help AI agents evaluate information, verify evidence and make decisions across business workflows without relying on a large language model for every step.
Available through an API, Bivia returns structured answers rather than generating output text for each decision. It handles five decision types: Choice, Boolean, Score, Set and Span. It works across text, images, documents, tables, charts, audio and video, and it can point to the supporting passage, image region or video timestamp behind a decision, along with a confidence score. When the evidence is insufficient, it can withhold a decision instead of answering without support.
The model offers three operating modes: Fast for routine decisions, Fast+think for more complex assessments, and Deep for tasks that need further verification. It can also be tuned to an organization’s own policies, procedures and past decisions. In one demonstration, Articul8 processed every frame of a 52-minute, 4K documentary filmed aboard the International Space Station in a single request.
Articul8 reports strong internal results, including near-perfect accuracy on selected energy and hardware design tasks after domain-specific training, and workflow completion of around 70% in one evaluation against about 30% for the alternative system tested. These are company-reported figures that have not been independently verified. Bivia is offered as a standalone API or as part of Articul8’s enterprise platform.
The launch reflects a wider shift in agent design: instead of routing every small judgment through a general chatbot such as ChatGPT or Claude, teams are adding specialized layers that decide, check and verify. Readers comparing the general-purpose options can see how they stack up in our ChatGPT vs Claude comparison, while Gemini offers another multimodal route for agent builders.
Source: MarTech AI
