The Problem
Amanda had a real problem to solve, but no way to build the technology to solve it. Autism and neurodivergent care in the U.S. had grown fast, from roughly 1,500 certified providers decades ago to over 60,000 today, but more than half of those providers have less than five years of experience. Ask ten clinicians to build a treatment plan for the same patient, and you'll get ten different approaches. There was no standardized way to guide a provider through intake, no consistent method for turning a patient's full history into a real treatment pathway, and no fast way to review medical records that often ran hundreds of pages long, with almost no time to read them properly.
Amanda had the clinical expertise and the vision. What she didn't have was a technical co-founder who had actually built a company's technology from the ground up before, and who could turn a healthcare problem into working software without burning through capital on the wrong approach before finding the right one.
The Solution
We started by proving the concept before spending real money building it. Robert built a working low-code prototype in LogicNets, giving Amanda something real to test and refine before a single line of production code was written. Before the full build began, Robert audited an outside development firm on Amanda's behalf and recommended paying for a dedicated planning phase upfront, a decision that avoided far more expensive rework later.
From there, Robert led the development team building the full platform, moving the engagement from a fixed-price model with built-in cost padding to a monthly retainer, giving Amanda more control over spend and better visibility into progress. The result is Knowetic's Care Navigation Module, which guides clinicians through an adaptive intake process that changes based on each patient's unique history, producing a tailored profile and treatment recommendations instead of a one-size-fits-all checklist.
Behind it sits KnowYeti, a purpose-built AI that clinicians can talk to directly, like a colleague with a photographic memory of every record they've ever read. KnowYeti can build timelines, summarize histories, and draft reports in minutes instead of hours, and because it's scoped only to the documents it's given, with no open internet access, it was built HIPAA-compliant from the ground up.
The Tech Stack
The art behind the infrastructure
Enterprise Cloud Architecture
Microsoft Azure + React
The entire platform runs on Azure, built as independent API components that call on each other, giving the system room to scale without becoming fragile.
Self-Hosted AI Model
Llama 3 on Azure
Unlike tools that send data to a third-party API, KnowYeti runs on a self-hosted model inside Knowetic's own Azure environment, trained exclusively on their own data.
KnowYeti, the Clinical AI Assistant
Custom LLM + OpenAI
Clinicians talk directly to KnowYeti about a patient's full record, like a colleague with a photographic memory, building timelines and reports in minutes instead of hours.
Adaptive Care Navigation Engine
Custom Decision-Tree Logic
A complex decision tree adapts every question to how the last one was answered, covering more than 120 possible report outcomes, and writes the final report automatically, saving clinicians 45 minutes per intake.
HIPAA-Compliant Infrastructure
Azure, Audited & Compliant
Built on HIPAA-compliant development practices from day one, with completed audits confirming the standard, not just claiming it.
EHR/EMR Integration API
(in progress)
Custom Built API
An API connecting Knowetic to any Electronic Health Record or Electronic Medical Record system is currently in development, extending the platform beyond its own walls.
The Results
Automation that became part of the DNA
$150,000+
Avoided in development costs, through smarter planning and a smarter engagement structure
120 customers
At peak adoption, including the University of Georgia and the University of Kentucky
45 minutes saved
Per clinician intake, thanks to auto-generated reports