HendrenAI
HendrenAI builds voice and workflow agents for pharmacies, health systems and pharmaceutical manufacturers. Its agents call payers and PBMs for prior authorization and benefits verification, answer inbound patient calls about refills, scheduling and navigation, run post discharge and chronic care outreach, and enter prescription data, writing structured results back into the pharmacy or hospital system. Pharmacy tools also cover inventory forecasting, diversion monitoring and medication reconciliation.
Routine interactions are handled automatically under an 85/15 model, and exceptions that need clinical judgment go to a pharmacist or clinician. Role based access, configurable guardrails and action histories support that split. Integrations with DRX, which reaches 400 pharmacies, and Liberty Pharmacy Software, which reaches about 3,000 independent pharmacies, are labeled live, and agents deploy into Epic, Oracle Millennium and athenahealth in about four weeks. OncoScope AI runs on the platform.
Founder and CEO John Boos is a pharmacist with an MBA, and the company works from an office in Montvale, New Jersey, with a team drawn from NYU Langone, Mount Sinai and Goldman Sachs. Pharmacy voice handling is priced as low as $0.10 per minute, while platform and implementation charges are quoted. A top ten US hospital is listed as a deal in negotiation.
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
Conversational language processing and agent workflows carry the pharmacy offering. Voice agents interpret inbound and outbound calls, extract details and turn conversations into structured actions around refills, benefits and prior authorization, and LLM driven agents with tool calling carry out those steps in connected systems. Prescription entry and forecasting take the product beyond a telephone interface.
Integration, workflow configuration, access control and operational routing carry value apart from any particular model, and the split between model output and configured logic in each workflow is not broken out. The calls, extraction and writeback that define the product still depend on the learned components.
Routine, repeatable interactions follow the 85/15 model, with about 85% handled automatically and exceptions such as unusual denials or decisions needing clinical judgment sent to a pharmacist or clinician. Configurable guardrails limit agent behavior, action histories support review, and the acceptable use policy places review and validation duties on qualified users.
Those are identifiable points of intervention rather than a bare claim of human oversight. The split does not set confidence cutoffs, an override procedure for every action or what happens when no reviewer responds, and the promise of no missed calls covers availability rather than correct clinical output.
The offering uses conversational AI, natural language processing and LLM driven agents to extract information and carry out workflow steps. The company's engineering roles are built around retrieval augmented generation, tool calling, LangChain, LangGraph, Azure OpenAI and AWS Bedrock, which describes the stack the team works on.
That gives a technical direction without fixing the model that serves a production pharmacy call. No release identifier, training dataset, update history or model change notice for customers is given, and a role requirement describes work to build or maintain a system rather than confirming that every listed technology sits in the delivered configuration.
Voice conversations, prescription information and payer responses pass through the agent layer before structured results reach the pharmacy or record system, and outbound calls carry patient details to payer and PBM representatives. The production language model and speech service providers are not listed in any product subprocessor register.
Engineering roles center on Azure OpenAI and AWS Bedrock, and IBM appears as a partner logo, but neither maps current patient data to a supplier and contract. The architecture is recognizable without a complete processing chain, so supplier retention, training rights, transcription services and the conditions for switching model providers sit outside the available service terms.
The impact figures are projections. A 200% projected ROI for health systems, $89B in potential labor savings across US health systems and 11 billion hours returned to patient care describe the market rather than measured customer results, and the pharmacy calculator estimates savings by comparing call hours and labor cost with the advertised voice rate.
The 85% automation figure has no named deployment, observation window or denominator. The top ten hospital is listed as a negotiation, the DRX and Liberty counts describe pharmacies those systems reach rather than customers, and engineering roles still describe work toward an MVP and first production customers. No evaluation reports completion rates, error rates or clinical benefit for the current agents.
Role based access, configurable guardrails and action histories control how agents act in healthcare workflows. The acceptable use policy requires an express written agreement before sensitive information is processed and leaves authorizations and consent with the customer, and engineering work runs on synthetic test data or data stripped of identifiers.
The October 2024 privacy notice covers the website only. It does not say how long patient audio, transcripts, prompts or outputs stay in the agent platform, whether they can train models or how deletion reaches suppliers, and no AI safety incident process is described. Test data practices say nothing about production conversations and prescriptions.
HIPAA and HITECH compliance are claimed across agent deployments, and sensitive data use requires a written agreement under the acceptable use policy. Role restrictions and audit trails support that posture in operation, and the platform lists HIPAA, HITECH and SOC 2 together as its compliance basis.
The privacy notice defines its covered service as the website and focuses on visitor and account information, so it does not govern prescription records or patient conversations the agents process. No business associate agreement, subcontractor terms or product scope can be read before contracting, which leaves the compliance claim clearer than the agreement that would govern real patient information.
The platform claims SOC 2 certification alongside HIPAA, HITECH, role based access control and complete action trails. The SOC 2 claim attaches to HendrenAI's own platform rather than only to a hosting provider, which makes it an operational assurance claim for the company itself.
The report type, examination period, auditor, covered systems and the report itself are not supplied, and the HIPAA and HITECH language covers separate obligations. The framework is named, while its current scope across voice agents, integration services, customer deployments and a distributed engineering team is not set out.
The current pharmacy offering covers prescription processing, refill communication, payer follow up, compliance monitoring and operational support. Exceptions needing clinical judgment go to pharmacists or clinicians, the acceptable use policy keeps decisions with qualified professionals, and no device clearance is claimed for these workflow tools.
Post discharge and chronic care outreach bring the agents closer to patients, and a separate Practice Intelligence Platform with symptom triage is planned for 2027, which would raise its own intended use questions when it launches. Medication reconciliation and administration verification would need their own intended use detail if used beyond operational checking.
Responsible integration, safety and accountability are stated values, and the product supports configurable guardrails and audit records. The company's quality engineering roles call for evaluation sets covering accuracy, hallucination, safety and guardrail adherence, plus regression testing of models and prompts, which describes governance work under way rather than a finished control framework.
No results compare speech recognition, task completion or error handling by language, accent, patient population or pharmacy setting, and no release approval criteria or outside review of model behavior are described. The stated commitment runs ahead of the evidence that it is being carried out.
Action histories and exception routing let pharmacists and clinicians inspect a workflow and challenge an agent's handling before a difficult case closes. The acceptable use policy makes users responsible for reviewing and validating output.
Those mechanisms support correction in operation, but no product accuracy guarantee, indemnity toward the customer or duty to repair harmful output is stated. The website terms carry disclaimers and a liability cap, yet they define the service as the website rather than the agent platform's contract. Patients and other affected parties have only general contact details, an email address and a toll free number, as a route to correction.
Pharmacy integrations are the most specific. DRX, with 400 pharmacies, and Liberty Pharmacy Software, with about 3,000 independent pharmacies, are both labeled live, and an October 2026 outbound workflow description adds structured writeback into DRX with Liberty and Red Sail as extensions of a universal API. Agents also deploy into Epic, Oracle Millennium, athenahealth and smaller EHRs in about four weeks.
Those statements name specific systems and an input and output workflow without a consistent deployment map or interface specification. Pharmacy counts measure potential reach rather than active connections, and two way operation in a named production customer environment is less clear than the stated capability.
Agents are configured to work inside existing EHR and pharmacy workflows, with integrations scoped to each engagement and EHR deployment taking about four weeks. Engineering roles center on cloud development across Azure and AWS, but no deployment guide fixes the hosting option or region for voice, model inference and prescription processing.
The Montvale, New Jersey office is an organizational location rather than a data residency commitment, and the website privacy notice permits processing elsewhere without governing the agent platform's data path. Tenant isolation, support access, backup locations and any option to run models inside the customer's own infrastructure are not described.
Pharmacy voice handling is priced as low as $0.10 per minute, or $6 per call hour. A calculator applies that rate to compare estimated AI cost with staff time, with an example of 600 monthly call hours costing $3,600.
That gives a price basis rather than a complete offer, since minimum volumes, telephony charges, implementation fees and eligibility for the lowest rate are not stated. Broader engagements start with one defined workflow and expand across functions through a sales process. No tier table or self service checkout fixes the platform's total cost, and the calculator's projected savings are not a contractual return.
Independent, retail, specialty and health system pharmacies are the main settings, with patient access workflows also aimed at pharmaceutical manufacturers and the wider platform reaching hospital clinical, administrative and revenue cycle teams. Tasks include refill inquiries, prior authorization follow up, prescription entry, benefits verification, inventory forecasting, diversion monitoring and post discharge outreach, and OncoScope AI runs on the same platform.
Pharmacy system partnerships open a route into those settings without measuring model performance there, and the hospital example remains an unnamed negotiation. Results for recognition accuracy, task completion and safety by specialty are not reported, and the 2027 practice triage product sits outside current coverage.
Alternatives to HendrenAI
Other products that do the same kind of work as HendrenAI, and what each one does.
Magical sells AI agents that do the repetitive operational work in healthcare organizations.
SpinSci sells AI agents for health system patient access.
Cadence Suite helps pharmaceutical manufacturers see where patients are getting stuck between a specialty prescription and ongoing treatment.
Lumistry builds pharmacy communication and operations software. Its Voice AI Assistant, introduced in April 2026, lets callers request refills, check prescription status and get answers about pharmacy hours or location in natural language.
AI voice agents for patient access and revenue cycle calls.
Latent Health sells pharmacy intelligence software that helps health systems get patients onto specialty, infusion and retail medications faster.
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 |
|---|---|---|---|---|
|
Voice handling as low as $0.10 per minute
$0.1 baseline
|
Pharmacy voice handling starts as low as $0.10 per minute. The wider platform is scoped around an initial workflow and expanded by agreement. | Sensitive data use requires a written agreement. No business associate agreement, module coverage or tier restriction is posted. | No standard implementation fee, minimum volume or telephony surcharge is listed. EHR deployment takes about four weeks. | Vendor Published |
The pharmacy cost calculator shows 600 monthly call hours at $3,600 using the advertised rate of $0.10 a minute. It sets no minimum purchase or full subscription price. Eligibility for the lowest rate, included services and charges for agents beyond voice are quoted in a sales discussion. Projected labor savings are estimates rather than guaranteed outcomes.