HealthConvos asks patients what is actually on their minds, then answers with recorded video performed by lifelike avatars. It offers no medical opinion — it listens, and it points back to the library or to the patient's care team.
Sixty-nine clips, each one a file with a script, a transcript, and a caption track. Your client's medical, legal, and regulatory reviewers watch them, mark them, and sign them — the same way they handle a video, a brochure, or any other finished asset.
What ships is exactly what they cleared, and it stays that way until you commission a change. Review gets a closed set to work through and a submission it recognizes.
The response set is finite and fixed. Reviewers see every clip in advance, sign them once, and that approval carries to every client who licenses the module. Familiar process, familiar paperwork, a defined endpoint.
Guardrails narrow the range of what it might say, but the review can only cover the process rather than the words. That's a workable model in plenty of categories. Patient-facing pharma is not usually one of them.
The questions do the listening. The agent's job is to get patients talking about what is actually worrying them — the thing they would not raise in a seven-minute appointment. It renders no medical opinion of any kind. Every answer routes to one of two places: back to the recorded library, or back to the patient's own care team.
That boundary is what makes an open conversation safe to put in front of patients. The agent can ask anything. It can only answer from the shelf.
The standard production is an actor performing copy written by someone who has never had the conversation. Patients recognize it in about four seconds, which is roughly how long they stay.
HealthConvos is built from composites informed by clinical practice — how people actually describe the week after diagnosis, what they are afraid to ask, the thing they only say on the second try. No individual's story, and no individual's record.
Then it is performed by lifelike avatars. Nobody's face, nobody's voice, nobody identifiable. Authenticity and confidentiality normally trade against each other; here the same production decision buys both.
What patients report is not that they found the information. It is that they felt heard — and that they left more confident than they arrived.
Patients think of the thing that matters on the drive home. By then the options are a handout, a portal login, or calling back — so most of them just don't ask.
Every session records what was asked and what was played back. Your client sees which concerns dominate, which arrive earlier than anyone expected, and which topics nobody opens at all.
That last one is the finding people underestimate. A section patients consistently skip is a piece of the education budget doing nothing, and until now there was no way to know.
The dashboard reports question topics and top questions in aggregate — no personal identifiers are requested, and no patient profiles are built. What you see is what people wanted to know, not who wanted to know it.
It also tells us what to build next. A question the library cannot answer is a gap with evidence behind it. Those gaps become the next commissioned clips, which is how a module gets better after it ships instead of aging from the day it launches.
Static patient education is finished the moment it is approved. This one compounds — every deployment sharpens the next version, and your client owns the insight that made it sharper.
A module is licensed to your account. You brand it, name the URL, and assign the managers who watch it. The client relationship stays where it already is.
Reviewed, captioned, and ready. No build phase, no production schedule to defend.
Logo, palette, and contact routes. The clinical content underneath stays exactly as approved.
Set the slug and point it wherever the campaign needs it to live.
Which questions came up, where people stopped, what nobody opened. Per client, kept separate.
Content is locked, and that is the point. Approved once, it stays approved for every client who licenses it. Editing would send each of them back through review individually — which is why changes come to us as commissioned work instead.
The complete module, live and running. Open it, click through it, take it into a client conversation this week.
The production system is disease-agnostic. New scripts and a persona spec go through the same pipeline that built diabetes — the stage structure, the persona arcs, and the review path all carry over unchanged.
It answers faster than a description does. Open the diabetes module, ask it the questions your client's patients ask, and see how it handles the ones that are hard to answer well.