Today we’re shining a spotlight on MetaLearner, (no relation)—an independent AI-powered data science platform. Focused on existing Enterprise Resource Planning (ERP) systems like SAP, Oracle, and others, MetaLearner helps businesses streamline their data retrieval and forecasting processes for improved operational performance. By leveraging Llama, MetaLearner enables enterprises to tap into powerful AI capabilities without the need for specialized expertise, making advanced analytics accessible to non-technical users.

Initially, MetaLearner adopted Llama 2 to manage workflow orchestration. Over time, that usage expanded to incorporate Llama into more advanced AI-driven functions, such as data summarization, online search, and even complex tasks like text-to-SQL. This evolution enhanced MetaLearner’s platform, providing users with real-time, actionable insights by integrating Llama’s capabilities directly into their proprietary tools.

“We were impressed by Llama’s leap in model performance and ability to efficiently manage AI workflows, setting it apart from other LLMs,” says Lim Ting Hui, Head of Technology at MetaLearner. “Llama’s strong privacy framework made it the ideal fit for our needs, and since then, we have continuously evolved our platform alongside Llama’s advancements.”

By integrating Llama 3, MetaLearner further democratized access to complex data science workflows. With the release of Llama 3.1, the team transitioned to both the 8B and 70B models to power the platform’s core functionalities. “On the day of Llama 3.1’s release, we upgraded our AI models within 30 minutes and immediately rolled it out to our customers,” Lim notes.

This swift move gave users a more robust data science experience, making it easier to derive insights, automate processes, and make informed decisions.

For the team at MetaLearner, Llama’s open source models provide a level of collaboration and transparency that would be difficult to achieve in a closed ecosystem. Open source platforms like Llama have helped the team gain access to tools that were once only available to big tech.

“Open source is the driving force behind the next generation of innovation,” says Rafael Nicolás Fermín, Head of Product at MetaLearner. “This model has empowered us to compete with tech giants by unlocking opportunities to leverage state-of-the-art technologies like Llama without the massive R&D budgets of large corporations. It’s a once-in-a-lifetime opportunity for us.”

Additionally, the collaborative nature of open source has inspired MetaLearner to innovate rapidly while giving back to the open source community. As Fermín explains, “While most of our AI workflow remains proprietary to MetaLearner, we actively contribute to the open source community as part of our commitment to give back.”

One example is Rust engine, which converts Python pandas data frames to Excel reports with a 95% reduction in compute time compared to the leading Python engine.

“We’ll continue sharing innovations that push the boundaries of performance and efficiency as we progress,” adds Fermín.

Like many AI adopters, MetaLearner encountered challenges when first integrating Llama. At first, the orchestration engine struggled to invoke the correct tools, but through persistent refinement of AI workflows and model fine-tuning, MetaLearner optimized its system for smooth, accurate performance. Today, Llama 3.1 is a key driver behind the company’s AI-powered solutions.

“With Llama, MetaLearner can efficiently provide critical access to advanced data science workflows and share cutting-edge AI capabilities to the heart of operational decision-making,” says Ben Levicki, AI Architect at Cleveland Cavaliers, one of MetaLearner’s early adopters.

To see how MetaLearner is revolutionizing the data science experience in ERP systems using Llama 3.1, check out the demo video.