SAN FRANCISCO, February 18, 2025 – Sama, developer of enterprise AI solutions with agile data labeling for model development and supervised fine-tuning, today announced the launch of Agentic Capture, a feedback framework for multi-modal agentic AI.
Built on Sama’s configurable capture platform, this new framework is capable of managing the types of multi-modal, complex use cases necessary for agentic AI across a variety of industries. Agentic Capture, built on Sama’s extensible data labeling platform, ensures consistent, accurate annotation through the use of high-quality rubrics and instructions.
The framework’s precision insights and transparent reporting provide a deeper understanding of the AI agent’s behaviors and its impact on model goals, and offers the human in the loop (HITL) validation necessary to effectively iterate on model performance.
AI agents are models that can act autonomously to make decisions, take actions and learn from interactions to help solve problems or achieve specific goals. The agentic AI market is expected to explode over the next five years, moving from a current value of USD 5.1 billion to USD 47.1 billion by 2030. The success of these agents will hinge on their ability to effectively mimic human behavior, which will require ingesting data from multiple modalities, including text, images, audio or video.
This new functionality is a step change from today’s current large language models (LLMs) which, while useful in augmenting human capabilities, are limited by their inability to process multiple modalities. Sama’s Agentic Capture framework makes it easier to integrate multiple types of data into the AI model, while providing the consistent human judgement needed to effectively utilize that data.
“Sama works with some of the leading AI model builders in the world, and that has placed us at the forefront of innovation. While LLMs have taken center stage for AI models, they are not without their limitations. We are in the early stages now of seeing the impact AI agents will have on the industries and companies we work with,” said Duncan Curtis, SVP of AI product and technology at Sama. “Agentic AI is on the precipice of exploding in capabilities and use cases, and our new framework is designed to make it easier to leverage data – whether text-based, images, audio or video – to enable AI agents to autonomously complete their tasks. We’re excited to continue working closely with our partners to build out a robust framework that meets the burgeoning needs of this technology.”
Use cases for AI agents run the gamut from human-like customer service in retail and insurance; to optimizing production in manufacturing; fraud detection in finance; resume screening, recruitment and training in human resources; analyzing consumer behaviors in marketing and much more. Sama’s data capture framework builds on the company’s decades of experience assessing and improving AI models.
For over 16 years, Sama’s proprietary HITL approach has consistently provided models with feedback from expert annotators, validating a model’s behavior and ensuring it is performing to standards. This feedback is woven throughout the entire model development process, including data creation, supervised fine-tuning, LLM optimization and ongoing model evaluation, ensuring clients can develop models in a more responsible way. Sama’s work is backed by SamaAssure™, the industry’s highest quality guarantee, which routinely delivers a 98% first batch acceptance rate.
