CareCompass
CareCompass is a patient navigation assistant from HealthAgents.ai, a Newport Beach, California company founded in 2024 by a team with more than 20 years in healthcare information technology. It is aimed at employee health plan networks, payer networks, regional healthcare organizations and retail healthcare providers, mainly midsize and large employers, to steer members to in network care and take routine tasks off staff.
The assistant recommends in network resources from a member's clinical needs, insurance details and preferences, books, reschedules and cancels visits, explains results and care plan updates in simpler terms, and answers questions about plan rules and balances, all in one conversation. In the company's mobile walkthrough an assistant named Jane offers a symptom check, care gaps or a results check, returns an impression on a described symptom, then shows provider profiles with distance, network and whether each clinician takes new patients before offering appointment times.
HealthAgents.ai calls its approach zero integration technology, saying the agents exchange data with EHR and payer systems without a separate integration project, and says its medical answers draw on premier content providers. It showed CareCompass at ViVE 2025. The site still invites pilot partners and names no customer, price or privacy terms, and its sign in page returns an error.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
The assistant is the product. A member types a question or a symptom, and the model turns open text into the next action, whether a plan rule, a provider search or an appointment, within one thread. Benefits questions, results explanations and care plan updates run through the same conversation, and scheduling and directory lookup exist to serve it.
Nothing about CareCompass works without the conversational layer, since no scheduling product or directory is sold on its own. That puts the language model at the center of every task the company describes, including the symptom impressions shown to members.
Booking, rescheduling and cancellation are presented as tasks the assistant completes on its own. When a member describes a symptom, the walkthrough shows an impression and possible causes returned straight to the member, followed by provider options, with no clinician in between.
Nothing says when a symptom concern is routed to a nurse, how staff can review an answer before it reaches a member, or how a scheduling action already sent to a provider is undone. For an assistant that offers a symptom check, the absent escalation rule is the central gap.
HealthAgents.ai describes an orchestration approach in which trained agents work across patient data sources and existing provider and payer systems. Recommendations draw on eligibility, plan network, clinical needs and where a member lives, and the output reaches members through a mobile chat with a symptom check, provider lookup and appointment flow. Medical answers are said to come from premier content providers.
The model family, which content sources feed medical responses, how retrieved content is verified and how a deployed model is changed are not named. The company also cites innovations and patents in healthcare and AI without identifying any.
Member input travels through several kinds of data. Clinical needs, patient records, coverage and preferences guide a member to a resource, and the agents are said to exchange information directly with record and payer sources, so a provider search, a symptom reply and a booking may each touch different systems.
The model provider is not named, and neither are any hosting, infrastructure or processing partners. Nothing separates the data used to find a provider from the data sent to generate a symptom impression, and no list of recipients or processing terms covers the member conversation.
A recorded walkthrough of one member journey, showing a symptom response, provider profiles and a conversation about open appointments, is the main material, along with a showing at ViVE 2025. The clinicians in the walkthrough are directory examples, not named customers.
The company says it aims to show better outcomes and satisfaction scores backed by real world data, and the pilot offer describes intended gains in engagement and resource use. No customer, baseline or outcome figure has been published for scheduling completion, answer accuracy or use of in network care.
The service touches sensitive information from several points in care. Members describe symptoms in chat, the assistant uses their insurance details, and it is marketed to explain clinical results and care plan updates drawn from provider and payer systems.
How identifiable details are minimized before an answer is generated, how long conversations are kept and whether they train any model are not explained. The privacy policy link on the site has no destination, so no document covers how member data is handled, and there is no described process for a safety incident.
Deployments span employer plans and provider networks, and the assistant takes in symptom descriptions and is marketed to pass on results and care plan updates, so its scope reaches well beyond directory lookups. It also answers questions about plan balances, which brings payment information into the same conversation as clinical details. The footer carries privacy and terms labels with no document behind them.
Business associate status, how a plan or provider authorizes use of member health information, any agreement tier and subcontractor commitments for the navigation service are all unaddressed.
The member interface handles provider search, booking and care communications, and the agents reach into existing plan and provider systems, so member data moves both inside the application and at every point where it calls another service.
No security architecture for those exchanges is described, nor operating controls such as organization roles or audit logging. There is no named independent attestation, certification or external test for the navigation service, no trust center, and the only contact listed is a general information email address.
Navigation, benefits questions and appointment management are the marketed uses. The product also offers a symptom check, and the walkthrough shows the assistant answering a member's concern with its impression and possible causes, which goes beyond choosing a provider or checking a balance. It is also marketed to simplify results and care plan updates for members.
No intended use boundary is drawn around the symptom workflow and no device status is given for it. Nothing requires a clinician to sign off before an impression reaches the member, who sees it in their own chat.
Provider choices are filtered by network, distance, specialty and whether a clinician is accepting new patients, inputs a member can see on each profile. The company says it applies AI suited to each problem and draws medical content from premier content providers for validated, evidence based output, without naming those sources.
No review process for model changes, no testing before a response workflow goes live and no comparison of results by language, insurance arrangement or population is described. The symptom impressions raise the sharpest question, since nothing explains how the assistant's reading of a complaint is checked or governed.
Members can open a provider profile and choose among appointment times, which gives them some control over where a recommendation leads. The same thread can carry a symptom impression, benefit details and clinical updates, so one conversation mixes logistics with health information.
There is no visible way to challenge an impression or send it to a clinician, to correct a booking made in error or to see the evidence behind an answer. No support channel for wrong answers, accuracy commitment or contractual remedy for harm from the assistant's output is offered, and the terms of use link on the site leads nowhere.
The pitch is zero integration technology: agents that exchange data with EHR and payer systems without a separate integration project, so provider discovery, scheduling and follow up can draw on existing systems. The walkthrough shows profile data and appointment openings in a single chat.
No record system, scheduling connector or exchange standard is named, and nothing confirms a production read or write against an outside EHR. How an organization connects its network data, what permissions the assistant needs and where a lookup ends and write access begins are left open.
Members use a mobile chat interface, and organizations buy through a pilot. The agents present provider profiles and openings and relay information from systems the plan or provider already runs, which defines the channels but not the infrastructure.
Where the agents run, where conversation history is stored and which regions are supported are not given. Tenant separation, backups, the location of model processing and any option to run inside an employer's or plan's own environment are not addressed either. The company's only stated location is its office on Newport Center Drive in Newport Beach.
Organizations reach HealthAgents.ai through a contact form, an email list and an invitation to discuss a pilot, and the target buyers are named: midsize and large employers, payers, regional healthcare organizations and retail healthcare providers. The company says skipping integration projects keeps costs low and time to value quick.
There is no pilot price, evaluation length or post pilot rate, and no unit of charge by member, appointment or conversation. What implementation includes is not described, so the cost of a plan or provider rollout cannot be estimated from the offer.
The intended settings are employee health plan networks, payer networks, regional healthcare organizations and retail healthcare, with midsize and large employers as the main buyers. In one conversation the assistant moves among a member's benefits, care gaps, clinical results and provider search, and the walkthrough follows one complaint from symptom to booking with network and availability as inputs.
No specialty, plan arrangement or region has a completed deployment named, and no results are reported by clinical service, geography or member population. Employer plan navigation is the most fully described use.
Alternatives to CareCompass
Other products that do the same kind of work as CareCompass, and what each one does.
Steer Health sells an AI workforce for patient access, clinical documentation and revenue cycle work at health systems, hospitals and medical groups, packaged as three products.
TalkToMedi sells MEDI, an AI front desk for family and specialty clinics.
Artera, formerly WELL Health (a US company unrelated to Canada's WELL Health Technologies), sells a patient communication platform, Artera Harmony, with AI agents across voice, text, email and web.
Menaela is an AI practice manager for solo practitioners and small private clinics, sold together with the practice management system it runs.
Tom is the AI primary care platform from Lumeris, the St. Louis and Cambridge, Massachusetts company that has supported value based primary care for health systems and physician groups since 2010 and runs the Essence Healthcare Medicare Advantage plan.
Patient access platform for health systems built around symptom triage and care navigation.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Not published
|
HealthAgents.ai invites healthcare organizations to discuss a pilot for member navigation and scheduling. No pilot price or unit of charge for an ongoing contract is published. | The assistant can use member information and communicate clinical updates. No business associate agreement position is stated, and the site's privacy link has no document behind it. | The company says its agents exchange information with provider and payer systems without a separate integration project. No implementation charge or onboarding scope is given. | Vendor Published |
The contact route is for healthcare organizations interested in a pilot. Pilot length, support terms and any rate after the pilot are not given.