
Technology
|
Supply Chain Automation
|
YC W26
|
Valuation:
Undisclosed

Last Updated:
March 24, 2026

Builds AI teammates for industrial distributors that automate quoting and order entry by parsing unstructured inputs (emails, calls, faxes, PDFs) using LLMs and NLP, then pushing clean structured orders into existing ERP systems.
Ventura has publicly detailed a modular "AI Skills" architecture for quoting and order entry automation, SOC 2 Type II compliance, and integrations with major ERP, CRM, email, and phone systems. They emphasize 95%+ automation rates, instant learning from user feedback, and rapid onboarding for industrial distributors. Their public messaging centers on expanding the library of AI skills and deepening system integrations.
GitHub activity from founder Swen Koller on "magentic," an open-source LLM framework, signals deep proprietary tooling around multi-model orchestration and agentic workflows. Conference and LinkedIn signals point toward expansion into predictive analytics (demand forecasting, pricing intelligence), automated bid screening, and agentic AI that runs continuous background workflows. There are also indicators of future support for new input modalities (images, handwritten documents) and industry-specific model fine-tuning on industrial/B2B catalog data.
AI-powered automation of quoting and order entry from unstructured inputs (emails, calls, faxes, PDFs) directly into ERP systems.
The AI reads messy emails, faxes, and phone calls from customers, figures out exactly what products they want, and creates a clean quote or order in the company's system—no human copy-pasting required.
It's like having a hyper-organized intern who can read your customer's chicken-scratch fax, cross-reference it against a 500,000-SKU catalog, and have a perfect quote ready before you've finished your coffee.
Modular AI Skills architecture enabling rapid deployment of new workflow-specific AI agents for distributor sales processes.
Instead of building one giant AI tool, Ventura creates small, specialized AI agents—like Lego blocks—that each handle a specific task and can be snapped together to automate an entire sales workflow.
It's like a Swiss Army knife where each blade is an AI specialist—one reads emails, one checks inventory, one writes quotes—and they all work together without bumping into each other.
Continuous learning and predictive analytics from distributor transaction data to optimize pricing, demand forecasting, and sales prioritization.
The AI learns from every quote, order, and customer interaction to predict which deals will close, what prices will win, and what products customers will need next—turning years of messy transaction data into a crystal ball.
It's like having a seasoned sales veteran's gut instinct—except it's powered by every transaction your company has ever processed and it never retires or forgets.
Swen Koller uniquely combines deep industrial distribution domain expertise (Harvard Business School MBA, BCG consulting, prior automation startup) with hands-on LLM engineering (author of the open-source magentic framework), allowing Ventura to build AI agents that understand both the messy reality of distributor workflows and the cutting edge of language model orchestration.