Sama Launches Agentic Capture, a Robust Data Capture Framework for Multi-Modal Agentic AI

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.

About Sama

Sama is a global leader in data annotation solutions for computer vision, generative AI and large language models. Our solutions minimize the risk of model failure and lower the total cost of ownership through an enterprise ready ML-powered platform and SamaIQ™, actionable data insights uncovered by proprietary algorithms and a highly skilled on-staff team of over 5,000 data experts. 40% of FAANG companies and other major Fortune 50 enterprises, including GM, Ford and Microsoft, trust Sama to help deliver industry-leading ML models.

Driven by a mission to expand opportunities for underserved individuals through the digital economy, Sama is a certified B-Corp and has helped more than 68,000 people lift themselves out of poverty. An MIT-led Randomized Controlled Trial has validated its training and employment program. For more information, visit www.sama.com.

Sama Media Contact:

press@samasource.org

LinkedIn Announcements:

Lisa:
Training AI agents is even more complex than training LLMs, which is why I'm excited to announce the launch of Agentic Capture, a framework built for training multi-modal AI agents.

Even in well-established procedural workflows, the actions taken by human agents are often cryptic and cluttered with superfluous information, making interpretation challenging. When issues arise, it's frequently unclear why they occurred or how to address them-leading to AI agents that fail to recover and adapt.

Agentic Capture transforms raw agent traces into interpretable steps, enabling effective oversight and meaningful improvements in AI agent performance. The solution includes:

• Strategic guidance that helps organizations develop a clear roadmap by defining the data and workflows necessary for effective feedback and alignment.

• A process-focused feedback system that enables continuous improvement based on detailed performance analysis.

• Proactive error detection that identifies model behavior gaps early and improves model understanding.

If you're in the process of training an AI agent - let's chat!

Wendy:
I am excited to announce the launch of @Sama's Agentic Capture, a feedback framework specifically designed for multi-modal agentic AI.

This framework addresses a critical gap in the AI industry: the ability to effectively build, monitor, and improve AI agents while maintaining comprehensive oversight.

The solution combines strategic guidance with a sophisticated monitoring interface, enabling organizations to track AI processes and implement meaningful improvements based on detailed performance analysis.

Our consultative approach, process-focused feedback system, and seamless integration with existing infrastructure help improve gaps in model understanding resulting in higher performance.