Understand the person
Clara builds a working picture of the person's health, circumstances, priorities, and concerns. This context helps make future conversations more useful.
A longitudinal AI health companion
Understand. Remember. Support. Over time.
Clara is designed to make every health conversation more useful than the last — by understanding the person, remembering what matters, and helping them move forward over time.
See Clara in action
A conversation that
continues over time.
Watch how Clara responds to a person's health questions while building context for future conversations.
More than answering today's question
Clara is designed around the reality that health management is ongoing.
Different conversations call for different kinds of support.
Remember selectively
Stable health information and meaningful long-term goals can become part of the person's continuing context, while temporary details do not need to become permanent memory.
Support the person
Clara supports the person in several ways, depending on what the situation calls for.
Collaborate on care
Clara can help turn priorities into practical self-management actions through an agreed Care Plan.
Bring in evidence when it matters 03:49
When current, authoritative, or medication-specific evidence would materially improve an answer, Clara can seek it out rather than relying only on general knowledge.
Prepare for clinical conversations 04:30
When a decision requires healthcare professional involvement, Clara can help the person prepare questions and organize a productive conversation with their clinician.
Support over time
Clara can follow up on what happened before, recognize progress, and continue from where the previous conversation left off. Over time, repeated interactions can create a richer understanding of the person's situation and help make future support more relevant.
Technical architecture
Built as a longitudinal AI system, not just a conversational interface.
Clara combines a conversational agent with persistent, structured health context, care planning, and grounded external research. The architecture is designed so that conversations can lead to durable updates and useful follow-up over time.
Conversational agent
A primary AI agent handles the ongoing relationship with the user. It interprets intent, reasons over available context, provides health guidance, supports goal coaching, and orchestrates actions such as memory updates, care planning, and research.
Longitudinal memory
Clara separates persistent health information from short-term working context. Health Profile captures relatively stable user facts, while Care Context captures what matters now. Care Plans and physician engagement records turn conversations into visible, actionable longitudinal state.
Grounded research
When external evidence can materially improve an answer, Clara can invoke a research layer that retrieves evidence using an appropriate research mode, including medication, clinical, patient-education, regulatory, and other evidence contexts.
From conversation to action
The frontend exposes the longitudinal state created through the agent workflow: Health Profile, Care Plans, physician appointments, and discussion questions. Update indicators make changes created by Clara visible to the user and connect the conversation to the rest of the application.
Research & background
From population-level targeting to individual support.
My research explores machine learning models for identifying future high-cost patients
Clara explores a complementary question: once someone may benefit from care management, how can longitudinal AI support make that care more useful at the individual level?
The research focuses on identifying who may benefit from intervention. Clara explores what comes next: understanding the person, remembering what matters, bringing in evidence when it helps, and supporting meaningful action over time.
Read the researchAcknowledgments
Built with help
from many places.
OpenAI / ChatGPT
Development support from concept and product design through architecture, implementation, iteration, and demo development. ChatGPT served as a co-developer throughout the project.
The open-source community
Built with and enabled by the broader open-source software community and the many tools, libraries, and contributors that make modern software development possible.
Google AI Studio
For access to AI models and development resources used during the project.
Anthropic Academy
For educational resources that helped build foundational knowledge in AI programming.