Document Processing · Marketing Personalization · Real-time Analytics
However, data labeling requires considerable resources and time. Manual labeling is costly and vulnerable to human errors.
Reported impact
Refuel.AI achieved measurable business impact on Databricks, including 86% Improvement in data label quality using a custom LLM; 3 Days to build a model from scratch with Databricks Training; Feedback cycles can be up to 100 times faster..
Access to high-quality, labeled data significantly improves model predictions and enhances context, boosting the performance of data analysis, ML, and AI applications. This produces more relevant results on search engine platforms and improved product recommendations on e-commerce sites. However, data labeling requires considerable resources and time. Manual labeling is costly and vulnerable to human errors. Human-in-the-loop methods only marginally speed up the process and increase accuracy. Large language models (LLMs) can enable vast improvements to this workflow. These challenges were the inspiration for startup Refuel.AI to build a platform for cleaned and labeled data that is powered by LLMs.
3 Days to build a model from scratch with Databricks Training
86% Improvement in data label quality using a custom LLM
Feedback cycles can be up to 100 times faster.
It outperforms trained human annotators (80.4%), GPT-3-5-turbo (81.3%), PaLM-2 (82.3%), and Claude (79.3%) across a benchmark of 15 text labeling data sets.
Launched in October 2023, the model was instruction-tuned on more than 5 billion tokens (comprising more than 2,500 unique tasks) on top of a Llama-v2-13b base model.
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Databricks Sample Jun 4, 2026 Red Hat But even in a highly data-driven marketing organization, teams trying to answer performance questions and find insights were slowed by fragmented information across dashboards, documents, spreadsheets and internal tools. $1.3M In annual productivity value Agent Bricks 
Google Cloud Sample Archived · Feb 15, 2026 17LIVE "All our content is on live, so to truly optimize the viewing experiences, we need to know what's happening and react fast. However, the data we analyzed didn't include what happened in our streaming rooms," explains… 17LIVE enhances its live streaming experiences by using Google AI to efficiently troubleshoot and create more… BigQuery · Gemini · Veo 
AWS Sample Jul 26, 2024 Daxte To develop AI models for its solution, the company fine-tuned open-source large language models such as Llama2 7b from Hugging Face with its own datasets and used Amazon SageMaker , a fully managed service that brings… Improving Performance Using AWS As of December 2023, LegalMap had a dataset of nearly 125 million documents… Amazon SageMaker 