Articul8 released Bivia, a model that makes the small yes-or-no calls inside AI agents, aiming to cut the cost of running them. The test results are the company's own.
Articul8, a company that builds AI for large businesses, has launched Bivia. It is a model that makes decisions for AI agents, the software helpers that carry out tasks on their own. The idea is simple. An agent should not need an expensive, powerful AI model for every small choice it makes along the way. Bivia is available now through an API, which is a way for developers to plug it into their own software.
Bivia can look at text, images, video, tables and charts, then make a call about what it sees. It gives a confidence score with each answer and points to the exact words, pixels or timestamps behind it. If the evidence is thin, it can abstain. That means it declines to guess, so the app can ask for more information or pass the decision to a person.
In a demo, Bivia went through every frame of a 52-minute, 4K documentary filmed from the International Space Station. It turned the footage into a searchable index, including moments that lasted only seconds, such as a glimpse of the aurora.
Articul8 cites a McKinsey report on AI in 2026. It says 40% of respondents from large companies were scaling AI agents, up from 27% a year earlier. About one in five said AI running costs, including token costs, were holding back their use of AI.
Bivia makes decisions without writing out answers the way a chatbot does. Developers can save the pricier generative models for harder work. Articul8 says customers including Hitachi Energy, CTC, AZing and Megazone are already seeing substantial savings. It gave no figures.
The numbers come from Articul8's own testing, so treat them as a starting point.
That last point matters. The company itself says those results show Bivia can be specialized, not how it handles data it has never seen.
Bivia is aimed at developers and companies building AI agents. Uses include checking evidence, routing tasks to the right tool, processing documents and confirming a task was finished. It can be tuned to a company's own policies and past decisions. It is sold on its own or as part of Articul8's larger enterprise platform.
The announcement does not give pricing. It also does not say how much money Bivia saves compared with other setups. Independent tests have not been published. A business weighing it would want to try it on its own tasks first.
If your business runs AI agents, ask whether every step really needs an expensive model, and test Bivia's claims yourself because the results are self-reported.
Otto, a new customer-tracking app for field sales teams, lets reps speak their notes instead of typing them. Price and launch details are not yet known.
Read more: Otto launches a sales app you talk to instead of typingTuring Award winner Richard Sutton says more computing power alone will not make AI truly smart. Big AI companies disagree, which matters for anyone betting on AI.
Read more: AI pioneer says piling on computing power won't make machines smartA viral ChatGPT prompt turns what the chatbot remembers about you into a room image. Posting it publicly could give scammers clues to target you.
Read more: Viral ChatGPT room image could hand scammers clues about you
