Scout InsurTech Interview with Endstation

Endstation builds custom AI automation systems that eliminate manual document and workflow bottlenecks for insurance, healthcare, and other regulated industries. Chris Luiz sat with Founder & CEO, Josh Tang, to learn more about how Endstation is automating the tedious.

Who are your clients?

“Endstation currently works primarily in the captive healthcare space, with openness to other lines of insurance. Right now we're doing a lot of work for a self-funded health plan client. We're open to other types of insurance.”

What does your product do?

“Rather than a single off-the-shelf product, Endstation builds customized AI tooling for individual clients. One core offering is a plan-document understanding engine, a system that ingests summary plan descriptions, provider manuals, and other regulatory documents to provide grounded, source-linked answers. Unlike a typical chatbot, the system includes a PDF viewer that highlights the exact source material behind each answer, allowing users to verify responses directly against the document. This tool is used by call center and customer success teams to answer member claims questions without looping in the claims department directly.

We also built a benefit comparison tool that ingests plan documents such as SPDs and summaries of benefits and quickly surfaces the details clients typically review when onboarding a new group, working toward cutting a process that used to take 60-90 minutes per plan down to under 10 minutes. More broadly, the company does generalized document processing work to extract and organize information across a range of use cases beyond chatbots.”

How much capital have you raised?

“Endstation is fully bootstrapped.”

Was the company born from within or outside the industry?

“The company was founded outside the insurance industry.”

What growth metrics have you accomplished over the last 12 months?

“Endstation grew revenue tenfold year over year. We have successfully delivered completed projects for four new clients, and received referrals to two new clients. Our number of senior engineers has doubled from two to four, with prior experience spanning the Federal Reserve Board of Governors, Microsoft, and large-scale distributed systems work.”

Within your domain, what is the current challenge that the industry is facing?

“There are three intertwined challenges. The first is signal-to-noise: AI has become such a buzzword over the last several years that the term has lost much of its meaning, and there's a temptation to assume an off-the-shelf tool can simply solve a problem without real investment in understanding it. The second challenge is data security and regulatory compliance, particularly around PHI and PII. The third challenge is underuse of historical data. I’ve found that many organizations collect extensive demographic and operational data but don't always take the next step of putting that historical data to work.”

How does Endstation take a unique approach to providing value?

“On the signal-to-noise problem, our approach is to build tooling grounded in the specific problem space, working closely with clients, listening to their actual needs, and iterating toward a well-defined set of correct answers rather than assuming a generic system will work out of the box.

On data security, we address the challenge by offering to host services on the client's own cloud or servers, ensuring no data is streamed to an outside AI company for training. Depending on the client's needs, that can mean zero data retention arrangements with major providers or running local, open-weight models in-house, sometimes fine-tuned specifically for that client.”

What inspired the team to start this company?

"The idea goes back to 2022, after encountering a 2021 research paper describing a reinforcement learning agent trained on thousands of hours of gameplay footage and connected to an early language model, allowing it to follow natural language instructions to complete tasks. At the time, I was working as a DoD contractor, running meetings and delivering ML models and AI systems, and recognized many of the same skills would translate to building something similar on a smaller scale. With no formal business background, purely an engineering foundation, I’ve spent the last three and a half years learning the business side of running Endstation.”

Can you share any goals for the next 12 months?

“The goal is to grow revenue roughly five times over the next year, a target I consider ambitious but achievable given it represents half the growth rate of the previous 12 months. Endstation's strategy has been to keep project scope narrow and deliver maximum value within a small surface area. Looking ahead, we want to build larger systems that reach further into underwriting and claims, building on the document processing and analysis work already underlying much of their claims-focused tooling.”

Scout InsurTech Thanks Its Partners