Customer-First AI: How innoviHealth® Ships Products That Win
The team expected a routine sales call. On the other end sat a very large company, the kind of client a young software business dreams about landing. What the client did not expect was to watch its own feedback become working software in the span of a single conversation. As their experts tried the product and asked for changes, the developers sat quietly on the line and made those changes in real time. Refresh the page. Is this what you had in mind? The client had rarely seen a vendor move like that.
About This Blog
The team expected a routine sales call. On the other end sat a very large company, the kind of client a young software business dreams about landing. What the client did not expect was to watch its own feedback become working software in the span of a single conversation. As their experts tried the product and asked for changes, the developers sat quietly on the line and made those changes in real time. Refresh the page. Is this what you had in mind? The client had rarely seen a vendor move like that.
In this episode of Fountain of Vitality, host LaMont Leavitt sits down with David Berky, Chief Innovation Officer at innoviHealth®, to trace how that instinct became a way of building. The conversation covers product speed, internal tooling, team culture in the age of AI, and the assistant now answering customer questions inside Find-A-Code™.
Ship Fast, Skip Long Cycles
Berky keeps coming back to one idea. When a customer names a need, the clock starts. He describes a company that does not worry much about long development cycles and instead pushes features to customers as soon as they are ready. That reputation followed the company into the market, where people came to see it as a fast mover and an early adopter.
The same thinking produced an internal data tool the team calls the panel. It imports data from many sources, organizes it so both staff and customers can make sense of it, and pushes it out to websites and APIs. The payoff is independence. Routine changes no longer wait in a developer queue, which frees the engineering team for work that actually needs them.
No Single Point of Failure
LaMont shares a frustration many people recognize. You call a company for help and hear that the one person who handles your issue is out for three weeks, so nothing can move. He calls that a clear sign of a process that needs documentation and automation. When a business locks up around one individual, the business is fragile.
Berky adds the practical fix. Critical functions get backups, and those backups get tested on purpose. The person who normally runs payroll steps aside for a week so the backup can run it, which surfaces gaps in the documentation before a real emergency does. LaMont pairs programmers on projects for the same reason, so work keeps moving when someone takes time off and nobody returns to a mountain of backlog.
AI as Support, Not Replacement
Fear showed up early. The customer service team heard stories of other companies handing their support desks to AI and worried they were next. Berky says leadership answered that fear directly. The goal was to hand the front line better tools, not to replace the people who talk to customers every day. Free them from routine work, and they have more time for the white glove service customers expect.
Developers felt a version of the same worry. The answer there reframed the whole role. Berky argues a developer's value was not in writing one more line of code but in understanding the customer, the system, and the business. Pull engineers deeper into the reasons behind the work, and a new AI tool becomes a source of excitement rather than a threat.
Cursor And The Time Collapse
The shift became concrete once the team adopted Cursor. Berky describes the early hesitation, the small tasks handed over as a test, then the growing trust as the results came back clean. Engineers even learned to have the AI review its own output with the right prompts, so their effectiveness grew in a way he calls closer to exponential than linear.
The numbers made the point. A feature that once meant several hours of work and a promise to circle back tomorrow now ships in fifteen or twenty minutes. Small changes that used to pile up under "I will get to it" started getting handled the same day, which shrank the backlog and gave developers room for the bigger projects they wanted.
Meet Aimee™ Inside Find-A-Code™
All of that led to a product. innoviHealth® had spent years building Find-A-Code™ into a large research library for medical billing and coding, licensing content from bodies like the American Medical Association and the American Hospital Association. The AI assistant, called Aimee™, sits on top of that organized data and answers questions directly instead of handing back a list of links to read.
Two things set Aimee™ apart, in Berky's telling. The data is authoritative and specific to medical billing and coding, not scraped from the open internet, and human subject matter experts shaped how answers get presented. The team has tracked a 99.8 percent satisfaction rate over recent months, and the website promises research that once took 47 minutes in a fraction of the time.
Test Every New Model
The last thread is discipline. New language models arrive almost monthly, so Berky builds systems that let the team swap models in and out and run a fixed set of test questions against each one. That way, they can tell if a new model gives better answers, equal answers, or worse ones before it touches a customer.
LaMont closes with a note for developers. Give management the data to make good calls, including token costs per answer, so the company builds the right product at the right price. Good decisions need good numbers, and in a market moving this fast, the teams that measure carefully are the ones that stay ahead.
Key Takeaways
Speed is a form of respect. Shipping features quickly tell customers they matter.
Long development cycles are a choice, not a law. Push work out as soon as it is ready.
Build internal tools so routine changes do not wait in a developer queue.
One person holding a whole process is a fragility, not a strength.
Test your backups on purpose before a real emergency forces the issue.
Pair people on critical work so nobody returns to an impossible backlog.
AI fear is usually a question about job security. Answer it with tools, not silence.
A developer's value lives in understanding the business, not in extra lines of code.
Feed an AI authoritative, specific data, and the quality of its answers climbs.
New models arrive constantly. Build a test harness so you can compare before you commit.
Follow the Fountain of Vitality podcast:
Website | YouTube | TikTok | Instagram | Facebook | Rumble | Email
Follow LaMont Leavitt:
LinkedIn | Twitter / X | innoviHealth®
Follow David Berky:
Related Blog
Duis mi velit, auctor vitae leo a, luctus congue dolor. Nullam at velit quis tortor malesuada ultrices vitae vitae lacus. Curabitur tortor purus, tempor in dignissim eget, convallis in lorem.
Subscribe Our Channel On Many Platform
Comments