← All organizations

Multimodal AI and digital agents

Adept

Adept develops multimodal AI models designed for digital agents that interpret computer screens. Its Fuyu-8B model accepts images and text and produces text, supporting research into computer control and visual understanding. Its intended capabilities include answering questions about interfaces, graphs and diagrams, and locating elements within screen images. Adept released Fuyu-8B as a smaller version of the model powering its product, with the public release intended for researchers who could adapt it to specific tasks.

Fuyu’s architecture feeds image patches directly into a decoder-only transformer through a linear projection, without a separate image encoder. It processes those patches as a sequence, using a special token to mark image row breaks. This lets the model accommodate different image resolutions without separate low- and high-resolution training stages. The released model is a base model: applications such as multimodal chat may require further fine-tuning.

In June 2024, Amazon entered an agreement to license Adept’s technology and hire its cofounders. That licensing and hiring arrangement marked a significant transition; it was not an outright acquisition of Adept. The company’s public model releases also include Persimmon-8B base and chat models.

1 talk

Newest first

Loading more talks and speakers…

1 speaker at AIE

Affiliations reflect their AIE event listings and recorded appearances, not necessarily current employment.

Company sources · checked 2026-11-09