MetaLearner
By MetaLearnerMetaLearner is an AI-powered platform that makes ERP systems smarter. We help manufacturers, retailers, and logistics companies unlock the full value of their ERP data by automating querying, cleaning, and feature engineering, while delivering advanced forecasts that factor in tariffs, weather, and macroeconomic shifts. Our proprietary AI agent architecture, spanning Database, Forecast, and Exogenous Agents, streamlines end-to-end forecasting and supply chain modeling. This allows teams to generate accurate, explainable forecasts directly from their ERP, replacing manual processes and siloed spreadsheets. Backed by Nvidia’s Inception program and leading advisors in ERP, supply chain optimization, and forecasting, we’ve already secured six-figure contracts and global POCs. MetaLearner transforms raw ERP data into actionable decisions that drive margins, resilience, and efficiency.
- Stage
- In production
- Industries
- Consumer goods
- Retail and e-commerce
- Logistics
- Deployment
- Cloud (SaaS)
Why it is worth it
1. Agentic Workflow Orchestration o Unlike black-box competitors, MetaLearner deploys a network of AI agents (Database, Forecast, Exogenous) that orchestrate the entire forecasting process end-to-end. o This allows real-time adaptability to disruptions and creates a defensible moat: the more workflows automated, the harder it is for others to replicate. 2. Data Ontology Advantage o MetaLearner maps messy ERP tables into a unified data ontology, enabling plug-and-play integration across SAP, Oracle, and NetSuite. o This ontology creates stickiness: once a customer’s ERP is structured through our system, switching costs rise dramatically while forecast accuracy compounds. 3. Accuracy + Explainability Together o Ensemble forecasting agents deliver top-tier accuracy while explainability agents show why forecasts change and what factors drive them. o Competitors force a trade-off between accuracy (complex, technical platforms) and simplicity (ERP-native tools). MetaLearner breaks that trade-off. 4. Minimal Technical Lift o Automated data prep and feature engineering mean minimal IT support and faster onboarding. o This opens advanced forecasting to mid-market manufacturers and retailers that lack large data science teams, expanding the market beyond what legacy tools serve.
- 1. Low Forecast Accuracy
- Most supply chain forecasts are inaccurate because they ignore external shocks (tariffs, weather, inflation) and rely on outdated statistical models or manual spreadsheets. MetaLearner solves this with ensemble forecasting agents that combine multiple advanced models and incorporate external data sources.
- 2. ERP Data Complexity and Siloes
- ERP systems like SAP and Oracle have hundreds of thousands of tables, making data extraction, cleaning, and feature engineering slow and error-prone. MetaLearner’s Database Agents automate this process, turning messy ERP data into usable inputs for forecasting and analysis.
- 3. Manual, Time-Consuming Processes
- Traditional demand planning requires endless Excel manipulation and siloed workflows, which are both inefficient and fragile. MetaLearner eliminates this with end-to-end agent orchestration, reducing onboarding and forecast generation from weeks to hours.
- 4. Lack of Explainability
- Most forecasting tools are black boxes, leaving supply chain teams with no insight into the drivers behind predictions. MetaLearner provides real-time explainability, showing why forecasts change and which factors are driving outcomes.
- 5. Slow Adaptation to Market Shocks
- Global supply chains are constantly hit by disruptions, but existing tools update too slowly. MetaLearner’s agents can dynamically integrate new data signals and update forecasts in real time, improving resilience and agility.
Opening the deck…
Page 1
Best customer profile
Our ideal customer profile includes CPG manufacturers, wholesalers, and retailers. These organizations manage a high volume of SKUs and transactions, making them an excellent fit for MetaLearner. They face daily challenges such as forecasting demand, managing inventory, and ensuring product availability across warehouses and distribution networks. Because our platform specializes in data extraction, forecasting, and optimization, we can directly improve their efficiency and bottom line. In addition, the frequent product rotations, promotions, and channel complexities in the CPG industry allow us to demonstrate value quickly and scale our solution across teams and regions.
Notable paying customers
- One of our most notable paying customers is Industrias San Miguel (ISM), a leading beverage manufacturer founded in Peru and now operating in multiple countries, including the Dominican Republic and Brazil. ISM produces soft drinks, bottled water, and fruit nectars, and is widely recognized for its commitment to quality and sustainability, holding international certifications such as ISO 9001 and ISO 14001. Our current contract with ISM is valued at USD 240,000, and we are actively working in three of their six operating countries, with a commitment already in place to expand to the remaining three. Specifically, we are supporting their executives and senior managers by streamlining data extraction and providing warehouse-level forecasting to optimize product availability across their distribution network.
A major milestone for us was starting with Industrias San Miguel (ISM) in a single country and then being invited to expand into two more, tripling our scope and doubling our contract size. Even as a young startup, we’ve rolled out multiple modules with them, including inventory management, forecasting, data extraction, and visualization. We are now executing robust optimization projects led by a team of PhDs from Singapore and China under the leadership of Melvyn Sim, which is laying a strong foundation for advanced supply chain decision-making. Beyond ISM, we’ve also worked with other CPG brands such as BIC in the Caribbean, demonstrating that our solution resonates across different types of companies and regions. We recently closed a deal with an industrial park zone to deploy our solution in manufacturing facilities, and with our current pipeline we aim to close 8–10 additional customers by year-end. On the funding side, MetaLearner has raised $2.3M to date, enabling us to accelerate product development and sales. From a technical perspective, we achieved a major breakthrough in collaboration with Melvyn Sim, a leader in Operations Research. Together, we’ve developed a new methodology that combines forecasting with robust optimization, enabling AI agents to design adaptive inventory policies that protect resilience without sacrificing cash flow. We are preparing to publish part of this work in INFORMS.