FollowMyDoctor.com was founded in 2018 with a single conviction: that artificial intelligence, properly engineered and carefully guarded, could make every clinician dramatically more effective — without ever replacing their judgment.
The idea began not in a boardroom, but at a clinical workstation — watching physicians spend more time navigating information silos than caring for patients.
In 2018, the founders of FollowMyDoctor.com — clinicians, engineers, and AI researchers affiliated with United Valuecare, Inc. in Chantilly, Virginia — identified a gap that no general-purpose technology had solved: the absence of a clinical AI platform designed specifically for the workflows, evidence standards, and regulatory constraints of licensed medical professionals.
Electronic Health Records had digitized medicine's data but not its reasoning. Search engines surfaced literature but couldn't interpret it in clinical context. General chatbots could answer trivia about diseases but could not participate meaningfully in a differential diagnosis or generate a compliant clinical note.
FollowMyDoctor.com was built to close that gap. Not as a chatbot. Not as a search engine. As a clinical intelligence platform — designed from day one to serve licensed professionals operating under HIPAA, CMS, and the full weight of clinical responsibility.
Headquartered in Chantilly, VA with a research office in Rockville, MD and a clinical AI lab in Boston, MA, we have spent six years building the infrastructure, the model library, and the guardrail architecture to make that vision real.
2018 — FollowMyDoctor.com incorporated as a division of United Valuecare, Inc. Initial research into clinical LLM architectures begins.
2020 — First internal prototype of the Medical AI Platform deployed to beta clinicians. HIPAA compliance framework established.
2022 — Partnership formalized with Deep Cognition Labs (DeepCog.ai) as the exclusive AI model engineering partner for the FMD platform.
2023 — AI Scribe and HIPAA Dialer launched. 100,000 clinician milestone reached. SOC 2 Type II achieved.
2024 — Patient Health Testing ($365/yr) launched. OpenMedLLM and OpenGenomeLLM open-source model families released on HuggingFace.
2026 — 300,000+ clinicians served. 31+ specialized AI model families. Agentic clinical reasoning systems in active development.
"To give every licensed clinician access to medical-grade AI — intelligence that reasons with evidence, documents with accuracy, and never mistakes confidence for competence." — FollowMyDoctor.com Mission Statement, 2018
The AI models powering FollowMyDoctor are not general-purpose models repurposed for medicine. They are purpose-built, domain-specialized systems engineered by Deep Cognition Labs against biomedical, clinical, and regulatory standards.
Every model is pre-trained on curated corpora of peer-reviewed biomedical literature, clinical guidelines (ACC/AHA, NCCN, USPSTF), FDA labeling, and de-identified clinical notes — not general web text.
Models are fine-tuned on structured clinical tasks: differential diagnosis, SOAP note generation, prior authorization drafting, genomic variant classification — each evaluated against task-specific gold-standard benchmarks.
Reinforcement Learning from Human Feedback (RLHF) pipelines use preference data collected from licensed clinicians — not crowd-sourced annotators — ensuring reward signals reflect genuine clinical quality standards.
For high-stakes queries, our platform uses Multi-LLM Uncertainty via Subset Ensemble (MUSE) — routing questions through a trusted model ensemble and surfacing calibrated confidence signals, not single-model outputs.
All models are evaluated against MedQA-USMLE, MedMCQA, PubMedQA, MIMIC-III clinical NLP tasks, and ClinVar variant classification accuracy — not general-purpose NLP leaderboards.
Our models undergo continuous monitoring against evolving guidelines, new drug approvals, and emerging literature — with systematic retraining cycles governed by a clinical review board.
Models are trained to decline requests outside their validated scope — diagnosing without sufficient clinical context, prescribing without a licensed prescriber, or asserting confidence on questions with insufficient evidentiary support — rather than producing plausible-sounding but unsupported outputs.
Every clinical claim is anchored to a specific source — a guideline, a drug label, a clinical trial. Outputs that cannot be grounded to a verifiable source are flagged or suppressed. Source citations accompany every clinical response.
The MUSE framework provides a calibrated confidence signal on every high-stakes output. Clinicians see not just an answer, but a quantified confidence level — with explicit prompting to seek specialist review on low-confidence outputs.
Patient data submitted via AI Scribe is processed in isolated, ephemeral, encrypted contexts. PHI is never used to train models without explicit BAA authorization. PHI boundaries are enforced at the model layer, not just the application layer.
FollowMyDoctor is architected as a clinical decision-support tool, never a decision-making replacement. Every AI output is framed as a recommendation for clinician review. The platform actively discourages AI output being used as a final clinical determination.
DeepCog.ai maintains a post-deployment adverse event monitoring pipeline that tracks clinician feedback flags, escalation patterns, and edge-case failures — documented, triaged by a clinical review board, and addressed in scheduled model updates.
The specialized medical AI models behind FollowMyDoctor are engineered exclusively by Deep Cognition Labs (DeepCog.ai) — a dedicated medical AI research organization and division of United Valuecare, Inc.
DeepCog.ai is not a general AI lab that also does healthcare. It is a specialized research organization whose entire model portfolio — 31+ model families — exists solely to advance AI in medicine, genomics, clinical reasoning, and biomedical research.
The partnership between FollowMyDoctor.com and Deep Cognition Labs is structural: DeepCog.ai engineers, validates, and maintains every AI model deployed on the FMD platform, while FollowMyDoctor provides the clinical workflows, compliance infrastructure, and end-user interfaces through which clinicians access those models.
Models available through the FMD platform include:
Chantilly, VA & Rockville, MD. A division of United Valuecare, Inc. Dedicated to building the world's most specialized medical and genomic large language models.
All models deployed on the FollowMyDoctor platform undergo DeepCog.ai's full clinical validation pipeline — domain-specific pre-training, supervised fine-tuning, RLHF with clinician feedback, MUSE uncertainty quantification, and continuous post-market surveillance.
Visit DeepCog.ai →Every product decision is evaluated through one lens: does this make a licensed clinician more effective without compromising their judgment or their patients' safety?
HIPAA compliance is not a checkbox — it is a design principle. PHI is isolated at the model layer, encrypted in transit and at rest, and never used for model training without explicit authorization.
We don't generate clinical outputs that can't be traced to a source. Peer-reviewed literature, FDA labeling, and established guidelines are the foundation of every clinical response our platform produces.
We publish our foundation models (OpenBioLLM, OpenMedLLM) as open-source contributions because we believe the best medical AI is built collectively, with full transparency.
Incorporated as a division of United Valuecare, Inc. in Chantilly, Virginia. Research into clinical LLM architecture begins, focused on the gap between general-purpose AI and the evidence standards of clinical medicine.
Core platform infrastructure built. HIPAA compliance framework established including PHI isolation at the model layer, BAA infrastructure, and encrypted data handling. First internal prototype deployed to beta clinicians.
Formal partnership with Deep Cognition Labs established. First specialized medical model families released. Platform opens to clinicians across primary care, emergency medicine, and hospital medicine.
AI Scribe — ambient clinical documentation powered by DeepCog.ai models — and HIPAA Dialer launch. 100,000 clinician milestone reached. SOC 2 Type II compliance certification achieved.
Patient Health Testing ($365/yr) launched, bringing clinical-grade lab interpretation to individual patients. OpenMedLLM and OpenGenomeLLM open-source model families released on HuggingFace.
Next-generation agentic systems in active development — moving beyond Q&A to multi-step clinical workflow automation, multimodal EHR integration, and autonomous care coordination. 300,000+ clinicians served.
Join 300,000+ clinicians who use FollowMyDoctor.com for AI-powered clinical decision support, documentation, and patient care — built on the most specialized medical AI models available.