Inference and verification

AI is finding applications in the workforce, media, and content creation.

At Ralla, we’re exploring the potential for three related AI technologies to expand how we interact in person.

Our goal is to use AI to synthesize shared content, provide visual interpretation of events, and to verify actions as complete. By combining these capabilities together, we will unlock new forms of interaction, gameplay, and shared experiences.

Today, more people are becoming familiar with the use of Large Language Models (LLMs) to mediate requests to translate between modalities (e.g. text->images) and interpret complex questions. LLMs are also an excellent tool for interpreting and synthesizing information from a specific corpus of information. Ralla’s strategy is to build structured tools to collect information from activity participants and use LLM interpretation to derive shared goals, insights, and propose action from those combined submissions. The result unlocks applications in collective decision-making, social discovery, and group organization.

Visual Language Models (VLM) represent a subset of the capabilities of LLMs optimized to interpret images and provide details assessment of their contents. Applications derived from VLMs are used increasingly to verify product delivery, in construction, and in the insurance industry. In combination with the telemetry provided by mobile devices, VLMs are becoming an important tool to help verify status and progress on projects. At Ralla, we’re using VLMs to unlock the ability for groups to work independently and complete tasks with out third party verification, relying instead on conformation provided by AI analysis of captured images.

Facial Recognition

Facial recognition uses Convolutional Neural Networks (a simpler form of deep learning) to distinguish between individual faces. Ralla is exploring applications that enable new forms of gameplay using opt-in systems to share personal recognition for the duration of an experience.