Hello Patient

Conversational agents are the product, and the company is candid about where the intelligence comes from. It states it does not build foundational or voice AI models but takes existing models and fine tunes them for healthcare practices and front office workflows, and describes rolling out a multi agent architecture to handle complex workflows reliably. That is an honest architectural disclosure most vendors avoid, and it does not diminish centrality: without the agents there is no product, only a phone line.

This is among the more autonomous deployments in the index: agents answer calls, book appointments, run intake, answer insurance questions, take refill requests, and handle billing outreach, with the company targeting a 100 percent answer rate and reporting practices moving from 20 to 30 percent of calls unanswered down to zero. The tasks are administrative rather than clinical, which contains the risk.

What is not published is the escalation path when a patient raises a clinical concern mid conversation, or the confidence threshold at which a human is brought in, and for a system running triage adjacent intake that boundary is the thing a buyer should nail down.

Unusually forthcoming for this category. The company states plainly that it builds on existing foundational and voice AI rather than its own models, fine tuning for healthcare front office use, and that it has been fully generative and conversational from the start rather than layering AI onto scripted flows. Its voice infrastructure partner is publicly named. What is not published is which model providers sit underneath, evaluation methodology, or containment and accuracy measurement.

One party in the chain is named and the rest are not, and where the disclosures sit is worth noting. The voice infrastructure partner is publicly named, which in this segment is a real contribution: a patient call typically traverses telephony, speech recognition, a reasoning model and speech synthesis, and most vendors here name none of those tiers, so identifying even one lets a buyer begin the diligence rather than stall at the first question.

Compliance is also stated directly on the main site rather than behind a sales conversation, and the difference matters more than it sounds, because a claim a prospective buyer can read before making contact is one they can weigh while comparing vendors rather than after committing time to a process. The company also states it builds on existing foundational models rather than its own, which tells a reader that third party model providers are in the path even though it does not say which.

What is absent is those names, plus any hosting arrangement, sub processor list, retention position for call recordings, and any statement on whether conversations inform model tuning. That last is the one to press given the stated fine tuning for healthcare use, since fine tuning implies a corpus and nothing says whose. Ask which model providers sit underneath, what the fine tuning corpus was, and recording retention.

Operational metrics are specific but vendor reported and early. The company reports powering more than 100,000 phone calls and 300,000 patient conversations within roughly its first year, around 20 percent more appointments booked, and 25 to 30 percent conversion when reaching out to referral patients to get them scheduled. Named customers span med spa, behavioral health, and outpatient groups. The company was founded in 2024, so there is no longitudinal evidence, and no independent evaluation of conversation quality, error rate, or patient experience was located.

The company states HIPAA compliance and SOC 2 Type 2 certification directly, and the surface is real: agents handle inbound and outbound conversations that necessarily involve identifiable patient information, insurance details, and clinical context around refills and follow up. Stating both commitments plainly on the main site, rather than behind a sales conversation, is better practice than most vendors in this category. No disclosure was located on call recording retention or whether conversations inform model tuning.

HIPAA compliance is stated publicly and prominently alongside SOC 2 Type 2. No explicit business associate agreement commitment language was located, which is the element that would move this to an A given the agents process protected health information in every conversation they handle.

SOC 2 Type 2 is stated explicitly, and the type is specified, which matters because Type 2 tests operating effectiveness over a period rather than control design at a point in time. Naming the type rather than saying SOC 2 generically is a small but meaningful precision. No trust center or downloadable attestation was located.

Patient communications and scheduling automation rather than a clinical product, so no FDA clearance applies and none is claimed. Nothing published maps the platform to any framework a practice is measured against, and the regulatory language used is superlative rather than specific, describing best in class privacy standards and healthcare's strictest security standards without naming a standard, certification or control. Two scope points deserve direct questions.

Triage is listed among the agent's functions, and an autonomous agent that takes symptom information and directs a patient toward a level of care sits nearer the clinical decision support boundary than booking does, with no published statement of what it will not assess or when it must escalate.

Separately, veterinary practices are served alongside human healthcare, which places part of the platform outside HIPAA entirely, so a buyer should confirm which regime and which contract terms govern their own instance.

No AI governance material was located. There is no responsible AI resource, no framework or certification, no model or training data documentation, and no description of how the agent is constrained beyond a general statement about privacy standards. The measurement gap is what makes this consequential rather than merely thin.

The company reports more than 400,000 patient conversations handled in under a year, which is real operating scale, but publishes no containment rate, no escalation rate, no accuracy figure and no error analysis against it. Competitors in this category publish at least one such number. Scale without a quality measure tells a buyer how often the agent spoke, not how often it was right.

Language support is offered and the voice can be tuned for accent, yet no performance breakdown by language or accent exists, which is the more sensitive question for a system whose failure mode is mishearing. Worth requesting containment and escalation rates from production deployments, and accuracy by language.

The company is unusually candid about what it is not, which is a form of disclosure this index credits wherever it appears. It states plainly that it builds on existing foundational and voice models rather than its own, fine tuning for healthcare front office use, and that it has been fully generative and conversational from the start rather than layering artificial intelligence onto scripted flows. Both statements are useful to a buyer.

The first sets expectations about where capability and risk actually originate, and admitting it costs a vendor the proprietary mystique most of this segment cultivates. The second is a real architectural distinction rather than positioning: a system generative from the outset and one retrofitted onto decision trees fail differently, because the retrofitted product is bounded by its script and breaks visibly at the edges while the generative one will attempt anything and fail fluently.

A buyer told which they have knows what to test for. Held at C because nothing is measured. No model provider is named, no evaluation methodology exists, and no containment, transfer or accuracy figure was located, with no warranty, indemnity or remediation commitment. For a front office agent handling refills and follow up, the containment rate and the behaviour on a clinically urgent caller are the two figures that matter and neither exists. Ask for containment and transfer rates, the escalation rule, and accuracy on medication related requests.

Integration is the precondition for the product working, since booking an appointment requires writing to the schedule. The company states it integrates with the EHR, practice management, and CRM systems healthcare organizations already run, with API integration into those systems described as the delivery mechanism. No specific named EHR integrations, certifications, or standards support were located, so breadth is asserted rather than enumerated.

A vendor hosted cloud service integrating through APIs into existing electronic health record, practice management and customer relationship systems. Nothing beyond that is published. No hosting provider is named, no cloud region or data residency commitment is offered, no tenant isolation model is described, and no customer hosted option exists.

For a voice product the omission is more consequential than it first appears, because a call crosses telephony, speech recognition and speech synthesis providers before reaching the model, and each of those is a place where audio and its transcript exist, yet no sub processors are disclosed. The company also serves veterinary practices alongside human healthcare on the same platform, which makes the separation between those environments a reasonable question rather than a pedantic one. Worth establishing where calls are processed and recordings stored, how long audio and transcripts are retained, which providers sit in the call path, and whether a region can be fixed in contract.

The thinnest commercial evidence in this category. No pricing, pricing metric or contract structure is published, which is normal here, but neither is the outcome evidence that competitors use to compensate. Results are described qualitatively, with more appointments booked and fewer no shows asserted rather than quantified, and customer testimony is enthusiastic but carries no figures.

What is published is scale and funding: more than 400,000 patient conversations handled in under a year, a 22.5 million dollar Series A led by a named institutional investor, and named customers including a mental health provider and a medical spa group. Volume and capital are useful signals about viability, but neither tells a buyer what the platform costs or what it returns. Competitors in this category publish attributed revenue figures, containment rates or a public return calculator. Worth requesting reference customers with quantified outcomes and establishing the pricing metric early, since per call and per location models diverge sharply as volume grows.

Broad across outpatient settings and explicitly multi specialty, with named or demonstrated use spanning urgent care, primary care, dermatology, orthopedics, ENT, dentistry, mental health, med spa, and veterinary practice. Target buyers are multi location medical groups and health systems as well as digital health companies. Inpatient and acute care are outside scope, and the veterinary inclusion is worth noting as a signal that the platform is built around conversation patterns rather than clinical domain.