Asha Health is a company that uses Artificial Intelligence to help Medical Practices transform to be AI-Native and enable them to create their own AI Clinic. Asha Health educates, guides and supports chronic condition patients consistently throughout their treatment process and generates an automatic increase in revenue for the practice. Asha was created by Former Leaders of Google and Brex who are focused on automating and streamlining administrative aspects of Care Delivery.
Co-founded by Akkshay Khoslaa (Former Google Product Manager) and Gokul Kumarresen (Former Growth Engineer at Brex) due to both having experienced Chronic Conditions personally, as well as from discussions with Health Operators.
Accepted into Y Combinator Fall 2024 Batch (YC F24)
Has raised $500K across 2 Rounds of Funding from Y Combinator and CoreNest.
Launched Platform to allow Medical Practices to Deploy Their Own White-Labeled AI Clinics to Address Chronic Care Gaps Beyond Traditional Visits.
Generative AI and conversational AI for chronic care management, hospital administration automation, and clinical support
Limited technical details available; inferred from product descriptions on YC page and company profiles
Asha Health is a free, open-source, AI-based health assistant utilizing Llama 3.1 which allows users to converse with the system via natural voice conversation, receive health information/advice, send emails, schedule appointments, etc. The user may activate Asha Health by stating "Hey AshA" and conversate hands-free using Web Speech API for speech recognition and Piper/WaveNet for natural speech synthesis.
To deploy Asha Health you need to have Node.js version 14+ installed, npm version 6+ installed, a Google Cloud account set up with the Gmail and Calendar APIs enabled, and OAuth 2.0 credentials created. Additionally, you must configure your environment to allow Piper TTS to run (espeak-ng and model files); and, when running the application locally, you must type npm run dev in your terminal/command prompt.
In order to enable Gmail API (read-only), Calendar API in Google Cloud Console, and OAuth 2.0 credentials for your application, you must enter CLIENT_ID and CLIENT_SECRET into .env.local file. Also, you need to configure authorized origins and redirect URI's for your local development server.
Your data security is based upon the permissions you grant to your Google account and how you are hosting your local copy of the application. The application uses standard OAuth scopes. There is no mention of additional encryption or compliance certifications. A review of the conversation history is stored locally.
Asha Health is currently open-source for self/local hosted use only. I did not find any documentation or guides regarding cloud service, SaaS offering, or production deployment.
Asha Health utilizes the Llama 3.1 model for NLP/NLU. The application processes the text in a separate thread than where the audio is processed for optimal performance. I also found a Colab notebook that you can utilize to train the Llama 3.1 model.
There is no native mobile app. The features of the application are designed to be responsive so you can access them from any device with a browser (mobile included) and you can access the features via voice commands through the browser Web Speech API.
You can fork the GitHub repository, create a new feature branch, and submit pull requests. I did not find any documentation or links to official support channels, forums, or paid support. However, community contributions are always welcome. Beginning of the Text
Asha Health is an open source, self-hosted AI health assistant that provides voice activated personal health management. It is unique in its ability to be customized by developers locally using their own AI for health tools; however, it does not have the enterprise features, cloud services, and official support like other health AI products do. Therefore, it would best fit as a hobbyist / personal project and not a commercial health care solution.
Great open source prototype for developers but not yet ready for production healthcare use cases.
Asha Health is an open source project hosted on GitHub (sajanpoudel/Asha---Health-AI) that includes a voice activated AI health assistant using Llama 3.1, Google integration and local deployment. There appears to be no commercial SaaS service available through ashahealth.com even though it references the website. Only intended for use by developers/users and has no enterprise features or compliance certifications.
Limited - Primary source is GitHub repository. No official commercial product documentation, pricing, or company information found. Website appears inactive/defunct. Possible discontinued project.
Built using React.js + TypeScript and Next.js Framework and uses Web Speech API for voice and Llama 3.1 AI Core and Piper/WaveNet TTS. Architecture is dual-threaded and provides separate text processing from audio processing to provide optimal performance.
Asha's single-developer GitHub repository (https://github.com/sajanpoudel/Asha---Health-AI). Contributions are encouraged through a standard fork/PR workflow. The local version of Asha will require that you configure your Google account as well as install the required Piper TTS model.
Voice wake-word activation (Hey Asha) is in place for managing Gmail, scheduling Google calendar events, keeping track of conversation history, turning on/off dark mode, and making sure it looks good on all devices. It can also have conversations about natural health topics and offer some emotional support.
It runs locally by running npm run dev. It will need you to be using Node.js 14+ and to have configured your Google account for OAuth as well as installed espeak-ng for Text To Speech. There is no documentation on how to deploy it into a Docker container or to the cloud.
Integration with Electronic Health Records (EHR) and Practice Management Systems (PMS) to get direct access to the data.
AI agents to send out weekly check-ins and support for patients with chronic conditions in between physician visits.
Offers AI enabled voice calling capabilities along with AI enabled SMS (text) messaging & AI based workflows for coordinating patient care
Manages an AI team of 14 + agents that include AI receptionist and AI insurance verification, allowing medical practices to operate their practice fully automated by using the Asha platform.
Medical practices are able to create customized AI based clinics to support specific medical specialties such as internal medicine, cardiology, pulmonology etc.
| Specialty | Use Case | AI Capability | Evidence Base |
|---|---|---|---|
| Internal Medicine | Chronic Condition Management | Weekly patient check-ins, education, accountability between physician visits | Targets 120M Americans with chronic conditions; fills care gaps in overburdened practices |
| Cardiology | Ongoing Patient Support | AI clinic coordination for heart condition patients; automated engagement | Implemented across cardiology practices; generates revenue on autopilot |
| Pulmonology | Respiratory Chronic Care | Regular monitoring and guidance for pulmonary patients | Supports specialty-specific AI clinics with minimal staff involvement |
| Urgent Care/ED | Practice Efficiency | AI Receptionist, Insurance Verification, operational automation | 14+ AI agents run entire practice workflows autonomously |
| Primary Care | Care Continuity | Bridges gap between isolated physician visits; consistent patient support | Y Combinator-backed platform; $3.9M funding for expansion |
| Validation Phase | Data Characteristics | Key Metrics | Minimum Standards |
|---|---|---|---|
| Internal Validation | Practice data from EHR/PMS integration across multiple specialties | Patient engagement effectiveness, autonomous operation reliability | pending |
| External Validation | Multi-practice deployment across Internal Medicine, Cardiology, Pulmonology | Revenue generation, care gap closure rates, patient retention | pending |
| Production Deployment | Live clinical operations with 14+ AI agents | Autopilot reliability, staff time savings, revenue per patient | Performance guarantees with remediation for baseline feature failures |
| Continuous Monitoring | De-identified aggregated usage data | AI model improvement metrics, partner feedback integration | Data used to enhance offerings and train models |
The use of de-identified, aggregated partner data is solely for AI training purposes and to improve the Asha services offered to partners.
A Business Associate Agreement will be entered into prior to exchanging PHI (Protected Health Information) and a clear protocol exists for how PHI will be exchanged.
Confidentiality of all confidential information will be maintained through a return or destruction process when the agreement is terminated.
Documentation regarding the rights and limitations of the use of the partner data in the terms of service (TOS) agreements is clearly stated.
Compliance with CCPA (California Consumer Privacy Act), CPRA (California Privacy Rights Act) and all other applicable laws/regs that pertain to protecting individual privacy.
All customer/end-client feedback collected by Asha may be used by Asha as part of its product development/enhancement process.
