Skip to main content
☩ AI Health Guide
Menu
For Consumers
Best AI Health Apps Mental Health Apps Is AI Therapy Safe?
For Clinicians
Best AI Medical Scribes How AI Scribes Work
Compare
BetterHelp vs Cerebral Nuance DAX vs Suki vs Freed
Guides
AI in Healthcare 2026 Best Mental Health Apps
Methodology About

Stephan Kulik

Editor-in-Chief, AI Health Guide

Role

I run editorial operations for AI Health Guide — independent reviews of AI health and medical software for patients and clinicians. Responsible for testing methodology, pricing verification, regulatory reporting, and publishing cadence.

What I am — and what I am not

I am not a clinician. I do not hold a medical licence and I do not give clinical advice.

All clinical content on this site is reviewed by a named licensed medical professional before publication. That reviewer's credentials and review date appear on every clinically-reviewed page. Where a page does not yet carry a clinician byline, the reviewer is still being onboarded — see our Medical Review Board page for the active roster.

My editorial role: product testing, pricing verification, regulatory reporting (FDA AI/ML-enabled device list, HIPAA/BAA availability, state telehealth regulation), and sourcing discipline.

Editorial focus areas

  • AI medical scribes (Nuance DAX, Abridge, DeepScribe, Freed, Suki, Heidi, Nabla, Ambience)
  • AI diagnostic platforms (PathAI, Viz.ai, Aidoc, Tempus)
  • Consumer mental-health, fitness, nutrition apps
  • Digital therapeutics and FDA-cleared AI
  • EHR integration depth (Epic, Oracle/Cerner, athenahealth, eClinicalWorks)

How I evaluate AI health tools

The process I run for every review on this site:

  1. Discover. Continuous review of the FDA AI/ML-enabled device list, KLAS Research, and industry trade press (Fierce Healthcare, Becker's Hospital Review, MobiHealthNews) for new entrants and changes to existing tools.
  2. Test. Hands-on signup, account-opening, and feature exercise on every consumer tool I can self-onboard to (free tiers, free trials, paid trials where the price is recoverable). For B2B clinical software that requires health-system credentials, I substitute structured interviews with deployed clinicians plus KLAS data and vendor demos. Each review states which path applied.
  3. Verify. Pricing checked against the vendor's current public pricing page on the date of publication. Clinical-evidence claims cite a specific peer-reviewed study with the PubMed / NEJM / JAMA URL. Regulatory claims cite the FDA, FTC, HHS, or state-AG page directly.
  4. Publish. Clinical content routes to the Medical Review Board for sign-off before publish. Editorial-only content (pricing, feature matrices, EHR integration lists) ships without clinical review but is marked as such.
  5. Refresh. Pricing re-verified quarterly. FDA clearance status checked monthly. Regulatory events (DOJ, FTC, HHS-OCR settlements affecting reviewed tools) reflected within 72 hours.

What I've tested hands-on

This is the tool list I've personally signed up for, paid for where required, and exercised. I separate it from "researched via KLAS / clinician interviews / vendor demo" because the audit signal is different. If you're trying to decide whether to trust a review, the distinction matters.

  • Consumer apps (self-signup): Headspace, Calm, Wysa, BetterHelp (intake-questionnaire + matching flow only — therapy not exercised), Noom, MyFitnessPal, Insight Timer, Talkspace, Teladoc.
  • B2B with free trial: Freed (full 7-day trial, multi-encounter), Heidi Health (free tier).
  • B2B without consumer access (structured interview + KLAS only): Abridge, Nuance DAX Copilot, Suki AI, DeepScribe, athenahealth, DrChrono, Viz.ai, Aidoc, Tempus AI, PathAI, Phreesia, Amwell, Annalise.ai, Glass Health, OpenEvidence, Regard, Fabric Health.

Where I haven't tested directly, the review's UE-6 hands-on section is explicit about it. I'd rather mark a section "vendor demo only" than imply hands-on I didn't do.

What you'll not find here

  • No paid placements. Vendors do not pay for inclusion or for ranking.
  • No vendor preview of reviews. Reviewed companies do not see drafts and do not edit copy.
  • No affiliate-driven ranking. Our affiliate commission rates do not influence the rank order in any guide or comparison. We publish the affiliate amount in /methodology/ where the operator has disclosed it.
  • No medical advice. Every clinical claim is reviewed by a licensed clinician but the site as a whole is informational, not a substitute for a clinician evaluating your specific case.

Editorial independence in practice

The most-asked question from clinicians I interview: "what's the commercial relationship between your site and the vendors you review?" The answer is the same for every vendor: affiliate where the vendor has a program; otherwise no commercial relationship. Rankings are independent of commission. See /editorial-independence/ for the full statement.

Other sites I operate

I operate a small portfolio of comparison sites across regulated verticals, all under Kulik Media UG (Berlin, Germany). The reason this matters for a YMYL author byline: the patterns I've developed for sourcing discipline and regulatory reporting on adjacent sites — financial-services, prediction-markets, energy, AI SaaS — are the same patterns applied here. If you want to verify how I handle disclosures on a different regulated vertical, the easiest reference is any of the sites below.

  • Financial: neobanks (EU and global), banks-for-crypto, predictorhq (CFTC-regulated US prediction markets).
  • Energy: yourwatts (6-region green-energy portfolio).
  • Health-adjacent: biohackeratlas (consumer biohacking devices), peptide-guide (research peptides), thenootropiclab (cognitive supplements).
  • AI / SaaS: bestautomationtools (AI agents and automation), certselect (professional certifications).

The full operator entity (Kulik Media UG) is registered in Berlin; see our imprint.

Corrections log

AI Health Guide is a new site (first public version 2026-04). Corrections will accumulate publicly. Every correction is recorded with the publish date of the original error, the date of correction, and the substance of the change. See our corrections policy for the procedure and the running log. If you spot something wrong, the fastest path is [email protected]; we acknowledge inside 48 hours.

Languages

Editorial work in English. Operator communications in German (Berlin entity). The site is single-locale English-only; we don't translate clinical content because nuance translation errors in regulated health text are a real safety risk and translation oversight isn't a discipline we've built editorial capacity for yet.

How to verify

Featured: BetterHelp Get Matched ↗