Inspiration
A few years, ago, we all worked at a company called Benefitter that set out to fix a broken healthcare system. It may not surprise you to learn that we did not succeed in that goal.
I like to joke, though, that—as a by product of that work—we helped the largest health insurance company in the country to make an extra billion dollars per year. It wasn’t really an accident. We were very serious about applying our skills and experience at the highest level to make the best bulk health insurance enrollment system the world had ever seen. But we had no idea it would move the needle so much, and I don’t think UnitedHealthCare did either. But when UnitedHealthCare acquired Benefitter and rolled out that beautifully designed enrollment system, it suddenly became clear how much the old system had been harming them.
In the years since that experience, we've come to learn that poorly designed and burdensome enrollment and onboarding processes adding friction and sludge to crucial economic transactions everywhere, and with the recent advancements in AI-powered document processing, the time has finally come to fix the problem.
What it does
Vento allows financial services professionals to configure a "Vento Box" to walk their clients through a streamlined enrollment and onboarding process. Generative AI allows clients to avoid a lot of the drudgery, repetitive busywork, and common errors, by automatically reviewing submitted documents, verifying their compliance with the requirements in real time, and extracting the necessary data from them. This allows us to eliminate a lot of form-filling, manual validation, and delays that are common during typical financial services onboarding.
How we built it
We evaluated every major AI service during our initial discovery work, and found that Google Vertex AI/Document AI was the most capable and easiest to work with. We paired Google's document processing tools with Bubble.io's low code development environment to allow for rapid prototyping as we consulted with dozens of financial services professionals to learn about their onboarding challenges. For the Google AI Hackathon, we've chosen a Vento Box that we created to solve problems endemic in Wealth Management onboarding processes.
Challenges we ran into
Google Document Cloud makes the ingestion of documents pretty quick and easy, which is a good thing because the onboarding problem in financial services is a problem of diffusion and variability. Within each segment, in each firm, there are many different enrollment and onboarding processes, and their requirements change over time, sometimes even from deal to deal. That's why they have resisted automation. AI is remarkably good at "similar but subtly different."
However, this is not just a document processing problem. A lot of the typical delays and errors that muck up onboarding are the result of human behavior issues, and addressing them requires careful attention and good design. Good design really moves the needle.
One of the main reason that attempts to rectify or automation dysfunctional processes fail is because they’re using a rigid “waterfall” approach, while a durable solution requires a mindset of continuous development and empowerment of subject matter experts. Silicon Valley figured this out a long time ago, and the Agile methodology is universal in the software world. It hasn’t trickled down to the rest of the business world yet, for the most part.
Accomplishments that we're proud of
Adding Google Gemini as way to easily interpret the JSON files extracted from ingested documents. It makes results more flexible
What we learned
Generative AI provides another level of flexibility when it comes to dealing with data that contains lots of variations. It also adds features for further querying into the info and give customers a better overall experience.
What's next for Vento Box for Wealth Management Onboarding
We have several more pilot projects in the works, and we'll be building lots of new AI-powered automations to streamline document ingestion for them:
- A firm that buys debt from banks, but that debt comes in the form of thousands of unlabeled, unorganized PDFs
- Banks that struggle with conflict between compliance & risk people who want to run everyone through the wringer and salespeople who want their clients to be put on the fast track
- A financial analytics company that works with large farms and struggles to guide farmers through collecting the information they need.
- A firm that provides mortgage brokers with information on how consumers can improve their credit scores is growing fast through channel partnerships, held back primarily because they can’t collect the information, documents, and authorizations from each new broker fast enough.
