Forus

The automation is the product rather than a feature layered on a services business. The company describes itself as built from scratch on agentic AI without legacy technical debt, and the workflow it automates is the one a practice would otherwise staff: generating and submitting the authorization, drafting the appeal, routing the prescription, and chasing the payer. Third party profiling describes it as AI native rather than an AI enabled hub or outsourced service, which is the meaningful distinction in this category.

The system acts rather than recommends: it completes and submits authorization forms and drafts appeal letters without a human composing them. The disclosed control is an internally built LLM judge that checks form answers against the patient's clinical records and internal guidelines to verify what is written is supported by the medical evidence, which is a real self checking mechanism and more than most vendors publish. What is not published is where a human must sign off before submission, so a buyer should establish the review gate in diligence rather than assume one.

Architecture is described at the level of technique rather than provenance. The company discusses an LLM judge for form verification and a retrieval augmented appeal generator, which is more specific than most vendors offer, but no underlying model provider, hosting arrangement, or data retention posture for model training is named in public materials. The technique disclosure earns credit; the absent provenance disclosure caps it.

Evidence is customer testimony and vendor reported throughput rather than controlled study. Named practitioners describe moving from roughly five worked authorizations a day to fifteen or thirty, and approvals shifting from months to days, and the company reports eliminating roughly thirteen hours a week of clinical prior authorization work. These are consistent across multiple named practices, which is better than anonymous claims, but none is measured against a stated baseline methodology, and approval rate improvement is not separated from staff throughput improvement.

PHI handling is explicit: the company states all protected health information is covered under a business associate agreement, and the platform extracts patient records to complete authorization forms, so the PHI surface is substantial and acknowledged rather than obscured. The LLM judge that verifies form claims against the underlying record is a safety control on the specific failure mode that matters here, which is an authorization submitted with unsupported clinical assertions.

The company states plainly that the platform is fully HIPAA compliant and that all PHI is protected under a business associate agreement, published on the main site rather than buried in a trust page request form. For a platform that is free to providers, the BAA commitment is the load bearing contractual term and it is stated without qualification or tiering.

SOC 2 Type 2 certification is stated publicly alongside the HIPAA and BAA commitment. Type 2 is the meaningful variant since it tests operating effectiveness over a period rather than control design at a point in time. No HITRUST certification, ISO 27001, or public trust center with downloadable artifacts is evidenced, which is what separates this from the top of the axis.

No FDA pathway applies. The platform automates administrative and reimbursement workflow and does not make or inform a diagnostic or treatment decision, so it falls outside Software as a Medical Device. Buyers should note the separate regulatory surface that does apply here is payer side: the CMS interoperability and prior authorization rule reshapes electronic authorization requirements over the coming years, and no public position on that timeline was located.

No public AI governance framework, model monitoring disclosure, or bias evaluation was located. This is a live concern rather than a formality in this category: a system that predicts denials and drafts clinically grounded appeals could systematically underserve patients whose records are thinner, and nothing published addresses whether that is measured.

The company states integration with all major or cloud based EHRs, and the workflow design depends on it: the physician writes the prescription normally and the platform picks it up at the point of prescribing. Notably, it also publishes the fallback for practices whose EHR does not support integration, which is manual record upload, an unusually candid disclosure of the degraded path. No specific EHR vendor certifications or marketplace listings were located, so depth per named system is unverified.

No public disclosure of hosting arrangement, cloud provider, data residency, or whether a private tenancy option exists. The platform is evidently cloud delivered given the lightweight setup claim, but nothing beyond that is stated.

The price to the provider is published and it is zero, stated repeatedly and without a trial qualifier, which is more transparent than the contact us posture that dominates this index. The reason it does not grade higher is that the monetization model is disclosed only in outline: revenue comes from manufacturers and the life sciences value chain, with the company stating it provides those firms insight into how medicines perform. A buyer should understand what prescription level data flows to manufacturers as a condition of the free platform, and that boundary is not published in detail.

Stated coverage is any drug, any insurance, and any pharmacy, spanning pharmacy benefit and buy and bill workflows, across practices from solo independents to health systems. Named customer specialties skew to the high authorization burden fields where the product has obvious pull: dermatology, gastroenterology, rheumatology. A partnership with the American Gastroenterological Association reinforces the specialty concentration rather than contradicting the breadth claim.