Vantel Wants to Be the Operating System for Commercial Insurance Brokerages

The Macro: Insurance Brokerages Are Running on Duct Tape

Commercial insurance is a $900 billion global market, and a shocking amount of it runs on email attachments and manual data entry. I don’t mean the big carriers like AIG or Chubb, which have their own legacy systems. I mean the brokerages. The middlemen who connect businesses with insurance carriers, negotiate coverage, manage renewals, and handle claims. There are thousands of them, and most operate on a technology stack that would be recognizable to someone from 2005.

A typical commercial insurance brokerage processes hundreds of applications, submissions, and renewals per month. Each one involves extracting data from PDFs, entering it into a CRM (often a spreadsheet), comparing quotes from multiple carriers, and tracking the status of everything via email threads. The average broker spends more time on administrative work than on actually advising clients.

The insurance technology space has attracted significant venture capital, but most of it has gone to two buckets: direct-to-consumer insurance companies like Lemonade and Root, and infrastructure for carriers like Guidewire and Duck Creek. The brokerage middle layer has been largely ignored by the startup world. Applied Epic and Vertafore are the incumbents, and both are legacy systems that brokers tolerate rather than love.

What changed is AI’s ability to process documents. Insurance runs on dense, semi-structured PDFs: applications, policy declarations, loss runs, supplemental questionnaires. Until recently, extracting data from these documents required either manual entry or fragile OCR systems that broke whenever a carrier changed their form layout. Modern language models can read these documents with high accuracy and extract structured data reliably. That unlocks a completely different approach to brokerage software.

The Micro: A Data Scientist and an NLP Researcher Walk Into an Insurance Brokerage

Vantel is building what they call an AI-native operating system for commercial insurance brokerages. The platform handles document processing, customer relationship management, submission workflows, and carrier communications. Instead of a broker manually reading a PDF application, entering data into a system, and emailing submissions to carriers, Vantel’s system processes the documents, extracts the relevant information, and manages the workflow.

Love Redin is the CEO. He’s a data scientist who has worked at venture-backed startups and McKinsey/QuantumBlack, served as a venture capital investor, and built software at Sweden’s largest independent fund manager. He also worked as a claims analyst at Scandinavia’s largest insurer. That last part is key. He has actually worked inside the insurance industry, which means he has seen the operational pain firsthand. He also completed Arctic Army Ranger training and has degrees in engineering physics and pure mathematics from KTH Royal Institute of Technology in Stockholm. That is a resume.

Ulme Wennberg is the CTO. He’s an ex-Microsoft software engineer with over 900 research citations in NLP. He dropped out of a machine learning PhD, holds dual degrees in Engineering Physics and Business Economics from KTH and the Stockholm School of Economics, and co-authored a prominent 2019 information extraction paper. He built conversational agents at Amazon Alexa. The NLP expertise is directly applicable here. Insurance document processing is fundamentally an information extraction problem.

They’re a four-person team, part of YC’s Winter 2025 batch. The combination of insurance domain knowledge and serious NLP research credentials is unusual. Most InsurTech founders have one or the other.

The competitive field includes Applied Epic and Vertafore as legacy incumbents, plus newer entrants like Indio Technologies (acquired by Applied), HawkSoft, and Agency Zoom. On the AI document processing side, Quandri and Fenris Digital do specific pieces of the workflow. Nobody has combined the full OS approach with modern AI document understanding in a way that’s gained meaningful traction among brokerages.

The Verdict

I think Vantel picked a smart market. Commercial insurance brokerages are high-revenue businesses running on terrible software. The willingness to pay is there. A brokerage that processes $10 million in annual premiums will happily pay $50,000 a year for software that saves them two full-time employees worth of administrative work. The unit economics should work.

The risk is integration complexity. Insurance brokerages don’t exist in isolation. They interact with dozens of carriers, each with their own submission portals, document formats, and communication preferences. Building an OS means supporting all of those connections, which is a massive surface area for a four-person team. Applied Epic has been at this for years and still doesn’t support every carrier workflow.

At 30 days, the metric is document processing accuracy. If Vantel can extract data from insurance PDFs with 95 percent or higher accuracy across multiple carrier formats, the product sells itself through demos. At 60 days, I’d want to see full workflow adoption. Are brokers using the CRM and submission tracking, or just the document processing? At 90 days, the question is whether brokerages are replacing their existing systems entirely or running Vantel alongside their current stack. Full replacement means the product is working. Side-by-side means it’s a feature, not a platform.