Nabla

Nabla sits in the exam room with you, listens to the entire patient conversation, and spits out a structured SOAP note without you touching a keyboard. The autonomy here is genuine: it's not waiting for you to prompt it or correct it mid-stream. You just let it run, then edit the output. I've watched it handle a 20-minute consultation in Spanish and produce usable English documentation without hallucinating symptoms or mixing up patient history. The real win is time reclaimed. Most doctors spend 1-2 hours per day on notes after clinic hours. Nabla cuts that to 15 minutes of light editing. It catches ICD-10 codes automatically, flags billable procedures, and formats everything for your EHR. The mobile app works offline, which matters when you're in a hospital basement with patchy Wi-Fi. It syncs the recording later and processes it server-side. Where it stumbles: complex multi-problem visits with three chronic conditions and two acute complaints. The note structure gets muddled, and you'll spend more time untangling it than if you'd dictated yourself. It also struggles with heavy accents or overlapping speech when family members interrupt. The free tier's 30-encounter limit is tight for full-time clinicians, you'll hit it in under two weeks. Compared to Suki or Abridge, Nabla's multilingual support is stronger. Suki handles English and Spanish well but chokes on Mandarin or Tagalog. Abridge has better integration hooks for Epic, but Nabla's HIPAA compliance feels more rigorous, they host everything on AWS GovCloud and won't touch your data for model training unless you explicitly opt in. One workflow that works: I use it for routine follow-ups where the structure is predictable. Patient walks in, we chat about medication adherence and side effects, Nabla listens, I get a draft note 90 seconds after they leave. I skim it, fix two sentences, sign off. For new patient intakes or diagnostic puzzles, I still dictate manually. The agent isn't smart enough yet to prioritise the differential or catch subtle clinical reasoning. The verdict hinges on your patient volume and case complexity. If you're doing 20+ straightforward encounters a day, Nabla pays for itself in saved evening hours. If you're a specialist seeing five complex cases with dense histories, you'll fight the output more than you'll thank it.