Case study · Healthcare · AI assistant
HIPAA-compliant AI assistant for a library of 1,000+ health summits, citing the exact video moment
A healthcare summit organizer wanted visitors to get trustworthy answers from thousands of hours of expert interviews, without searching through them. With Ethora they launched a context-aware AI assistant on their WordPress site that answers only from their own content and links every answer to the right moment in the right video.
By the numbers
The project at a glance
1,000+
health summits of expert interviews made searchable by the AI assistant
≤ 30 s
target precision for the timestamp each answer links to
3
scopes: answers limited to the current video, the current summit or the whole library
$0.0007
approximate model cost per answer, on a 1 vCPU / 4 GB server
Overview
Thousands of hours of expert talks, one question away
The customer is DrTalks, one of the most prominent organizers of online healthcare conferences and summits, bringing leading medical experts to hundreds of thousands of health-conscious viewers. Each summit is a series of long-form video interviews on a topic such as heart disease, cancer or fasting.
The challenge. The answer a viewer needs usually exists, somewhere in hours of video. Search could find a talk, not the two minutes inside it that answer the question. For health content the bar is higher still: answers have to come from the experts on the platform, not from the open internet, and visitors’ questions have to be handled privately.
The approach. A retrieval-augmented (RAG) assistant that indexes the interview transcripts with their timestamps, retrieves the passages relevant to a question, writes a short answer from those passages only, and links to the moment in the video where the expert says it. It sits on the existing WordPress site as an embeddable chat widget.
Requirements
What the assistant had to do, and how it does it
Each requirement is tagged by what comes from Ethora’s AI stack and what was built for this content library.
Answer from the summit library, not the open internet
Interview transcripts and video metadata are ingested into a vector index. For each question the assistant retrieves the most relevant passages and is instructed to answer from that context only, in a few sentences, direct answer first. When the library does not cover a question, it says so with a fixed reply instead of improvising.
Cite the exact page, author, video and moment
Transcripts are split into paragraph chunks, and each chunk keeps its timestamp as metadata. An answer ends with a link to the video page at that second (#t=…s), so the viewer lands on the expert making the point, with a 30-second precision target.
Context-aware on every page
The widget reads where the visitor is. On a video page retrieval is filtered to that video, on a summit page to that summit, elsewhere to the whole library, so a heart-disease visitor does not get cancer content. The greeting changes with the context, for example “Hi, I’m the [summit] AI, ask me anything”.
Help visitors ask a good first question
Suggested questions are generated from the page the visitor is on (the video’s topics, the summit’s talks) and from the conversation so far, so the first tap already returns a grounded answer.
Privacy by design
Sessions are anonymous: no login and no identity are sent with a question. Conversation history lives in the visitor’s browser and is cleared automatically, and the server does not keep chat transcripts. The assistant answers only from published educational content, not patient records.
Fit the existing WordPress site and paywall
The assistant is an embeddable JavaScript chat widget added to the WordPress theme, with a full-screen layout on mobile. It works for visitors before the paywall, answering from transcripts, and for members on paid pages.
Keep support questions out of the medical answers
Questions about accounts, refunds or access are recognised and routed to the help centre instead of being answered from the interview library.
Cheap enough to leave on for every visitor
Roughly $0.0007 in model cost per answer, served from a single 1 vCPU / 4 GB server. Adding a video takes under a minute of indexing per transcript.
Architecture
A small RAG stack beside an existing website
No new platform for the content team: transcripts in, a widget on the site, one small server in between. The same pattern is available today as Ethora’s embeddable AI agent with a documents indexer, and can run on a self-hosted LLM when data must stay in your infrastructure.
Where this stack runs
“Ethora helped us deliver our AI agent solution, letting subscribers access context-specific insights from our vast library of expert content. Their LLM and chat-bot expertise is an excellent match for our AI strategy.”
Bret GregoryCEO, DrTalks
The outcome
An expert in the room, available to every viewer
The assistant soft-launched in spring 2024 on the summit pages and was handed over to run on the customer’s own infrastructure. A 2025 phase added expert, summit and topic context that follows the visitor across pages, and a full-screen mobile layout. Viewers get a short answer and the exact moment to watch, instead of a list of hour-long videos.
The pattern fits any content-rich health or education business: course libraries, webinar archives, clinical guidelines and podcast catalogues.
FAQ
Can the assistant answer only from our own content?
Yes. Retrieval-augmented generation retrieves passages from your library and instructs the model to answer from them only, with a fixed reply when the library does not cover the question.
Can answers be limited to the page the visitor is on?
Yes. Metadata filters scope retrieval to the current video, collection or the whole library, based on the page the widget is embedded in.
Is Ethora HIPAA and SOC 2 compliant?
Ethora is HIPAA and SOC 2 compatible. Compliance for a specific project depends on its configuration and hosting: for protected health information use dedicated or self-hosted deployment and a model provider covered by your agreements, under an Enterprise agreement with a BAA.
Can we add this to a WordPress site?
Yes. Use the WordPress AI agent plugin or the embeddable AI agent script, and point the agent at your documents or site content.
More case studies
Insurance: in-app chat for a workers’ comp health insurance provider (HIPAA compliant)
Atom Advantage provides workers’ compensation and health insurance solutions in the United States.
On Ethora they built Atom Connect: secure, case-based chat between injured workers, nurses and case managers, embedded in their claims portal and a white-label mobile app. HIPAA and SOC 2 compatible.
Less manual work for case managers and a better experience for injured workers.
- Chat embedded in the case managers’ web portal
- White-label iOS and Android app for injured workers
- One secure room per claim, created automatically by API
AI agents on customer websites: support, intake and teaching assistants
Clinics, schools and small businesses build AI agents on Ethora themselves and put them on their own websites, often on the day they sign up.
2 live examples, from a US mental health provider to a university pharmacology course.
- WordPress plugin or one script tag
- Answers from the customer’s own pages and documents
- Self-serve, no developer needed
Healthcare: self-hosted patient messaging for a multi-clinic digital health platform in Canada
VITALL is a Canadian digital health platform used by hospitals, clinics and private practices.
Ethora runs as a self-hosted chat server in VITALL’s own AWS account in Canada, with one isolated tenant per clinic, chat in the practitioner portal and in native iOS, Android and React Native patient apps.
Users, rooms and clinics are provisioned by VITALL’s backend through the Ethora API.
- Self-hosted in the customer’s AWS account in Canada
- One isolated tenant per clinic, created by API
- Web portal, iOS, Android and React Native SDKs
Communities: fan chat app with second-screen rooms and digital collectibles
A cross-platform social app uniting the automotive community around racing events.
Fan rooms for teams and drivers, second-screen events, and collectibles handed out and traded through chat bots.
- Cross-platform iOS and Android app
- Fan rooms and second-screen events
- Collectibles distributed in chat
Energy: dedicated chat server, AI-agent rooms and a React Native app for a procurement platform
Preshent is a US platform for energy-project procurement with AI-driven decision support.
Ethora provides the conversation layer: a chat room per project workspace shared by people and the AI agent, a dedicated chat server in Preshent’s own cloud, push notifications and a React Native app.
- 1,130 chat rooms, one per workspace
- AI agent as a participant in every room
- Dedicated chat server, 99.99%+ uptime
Build an AI assistant on your own content
Video libraries, courses, documentation or a whole website: Ethora indexes it, answers from it with citations, and embeds on your site in minutes.
