Silna
Silna runs prior authorizations, benefit checks and insurance monitoring for specialty provider groups on what it calls a Care Readiness Platform. It tracks upcoming and expiring authorizations, gathers supporting documents and follows each request through the payer's decision. Predictive Document Intelligence, launched in January 2026, reads clinical and administrative paperwork against each payer's requirements and flags missing information before submission. Browser agents file requests in payer portals, with rules that route selected submissions to Silna specialists for review, and a phone workflow calls insurers about benefits and authorization status. Benefit checks cover coverage, accumulators, authorization requirements and visit limits, and monitoring rechecks patients for active, lost or new plans.
The Interoperability Agent pulls request details from a practice's record or intake system and writes results back, configured to the practice's own process and tested before launch, with WebPT and Clinicient among the named systems and a REST API and bulk file exchange for others. At AnswersNow, a virtual autism therapy provider, submission time fell from 14 days to 3 and benefit checks from 72 hours to 19. Customers include Behavior One Autism Solutions, Autism Learning Partners, Elite Alliance Physical Therapy, Allergy & ENT Associates and HealthPro Heritage, and the company works with more than 1,000 payers across 50 states.
Jeffrey Morelli, Pavel Asparouhov and Sagar Jajoo founded Silna in New York in 2023, and Accel and Bain Capital Ventures led its $22 million Series A in March 2025. Quotes depend on authorization volume and modules.
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
Models do the reading and the clicking. Predictive Document Intelligence interprets clinical and administrative paperwork against each payer's requirements, browser agents use the extracted data to complete portal submissions, and phone agents handle certain insurer calls. Those are central functions in prior authorization and benefit verification.
Around them sits an administrative service that predates the document model: due date tracking, results storage and a team of operations specialists who review and correct work. Silna ran the workflow with that service before adding document intelligence in January 2026, and today's buyers get both, so the models are the engine of core modules rather than the whole product.
Prior authorization agents can submit on their own, with rules that send selected cases to specialist review, quality checks that flag a session that retries or backtracks, and tools that let specialists correct completed work. The benefits side is different. The homepage promises fast patient intake and 100% accurate quotes, the benefit verification page describes automatic checking, and no human approval is required before a quote reaches the practice or patient.
The review controls built for authorizations do not reach benefit estimates, and the terms disclaim any error free warranty. That leaves an absolute accuracy claim on the one output no person checks before it is used.
Separate models and validation stages handle payer portal tasks, and the engineering blog says Anthropic's Opus 4.1 ran the first production portal workflow. The team compares models in a test environment and can route a share of live requests to a candidate model, while the phone system runs its own chain of speech recognition, language interpretation and speech synthesis. That shows models chosen per task and per portal, with real version detail for one workflow. A full roster of the models live across all deployments, the training data behind the document model and a release history that would tell a practice when a model changed are not given.
The engineering team names Anthropic in portal automation and describes evaluations run on Anthropic and OpenAI models, and it names Deepgram for speech recognition and Twilio for calls in the phone workflow. That establishes the main model and voice suppliers feature by feature. It is not a complete subprocessor list for benefit and authorization work.
The service privacy policy puts confidentiality duties on providers, and Vanta supports compliance monitoring, but nothing maps which protected data each model supplier receives or sets the retention and training terms on each transfer, so the names come from engineering posts rather than a contractual list.
The October 2025 AnswersNow case study separates submission time from the insurer's turnaround. The virtual autism therapy provider went from 14 days to 3 to submit an authorization, from 29 days to under 7 on average for the full turnaround, and from 72 hours to 19 for benefit checks. Those are dated before and after figures from a named deployment. They predate the January 2026 document intelligence product and describe the broader service, not a separate effect of the newer model. Other figures carry less: more than 250,000 patients supported and a first place ranking for prior authorization on G2 at about 4.9 out of 5.
Under the July 2026 privacy policy, shared information is used to provide the service unless the customer consents otherwise, and anonymized and aggregated usage data may be used more broadly. Retention is described by factors rather than fixed periods for recordings, clinical documents or model inputs, and the phone engineering posts describe collecting insurance call recordings to develop that workflow.
In payer portals, the software is bounded in where it can send information and its actions are checked for inconsistencies. How clinical data is deidentified for development, when model suppliers delete inputs and how a customer is told about a harmful output are not set out.
The privacy policy covers the patient information processed for benefit checks and authorizations and describes transmission to payers and customer EHRs. The homepage carries a HIPAA badge beside the line that patient confidentiality is the company's priority, and the platform holds SOC 2 Type II.
Neither the terms nor the privacy policy identify Silna as a business associate for any service or offer an agreement for protected information, and no tier or contract form says how the AI functions and downstream suppliers such as Deepgram and Twilio are covered. The service policy describes handling in general terms, while the agreement and its scope are absent.
The site shows a SOC 2 Type II mark and says compliance runs through Vanta. Product controls include an audit record of every synchronization, tracing of the information used in a submission back to its source documents, and managed, isolated browser sessions for payer workflows.
The auditor's report and audit period are not presented, and there is no public portal where a customer can retrieve the current SOC 2 artifact or confirm whether the phone agents and portal automation were inside the assessed scope. The certification type is named, while its scope stays inside the report.
The platform works in the provider office before billing. It gathers supporting material, checks it against payer requirements and follows authorization and eligibility status with the insurer, without deciding coverage or making a clinical determination; the payer or its reviewer decides, and the treating provider supplies the documentation. The product is described around insurance administration and prior authorization, including how the preparatory document checks are used. No medical device authorization is claimed, and no treatment or diagnosis indication is offered for the document validation functions, which matches what the product does.
Automation is tested before it is switched on for a payer workflow, and a share of completed submissions is sampled for specialist review, with reviewers recording each error as minor or major and correcting the submission. The engineering team uses those errors to test changes and can compare models side by side in production. That is a named process for deciding which tasks can be automated and for watching the choice afterward. It reports no outcome or error comparison by patient demographics, language or coverage type, and no independent audit of model behavior across the 1,000 payers and their portals.
Specialists use a quality review interface to examine submissions, flag errors and correct the work, and the record sync keeps an audit log. The platform terms offer the service as is and disclaim any error free warranty, even though the homepage promises 100% accurate quotes. The privacy policy lets users ask for corrections to personal information, separate from deleting patient data. Nothing obliges Silna to correct a mistaken eligibility result within a set time, reimburse a practice for costs from a wrong quote, or let a patient appeal an authorization outcome to the company.
The Interoperability Agent pulls request details from a practice application and writes results back under a procedure agreed with the customer, tested on sample data before activation and logged on every sync. Physical therapy material names WebPT and Clinicient as supported systems, and REST access and bulk file exchange cover other setups. That shows the direction of travel and the systems involved. No named production installation ties the full read and write workflow to WebPT, Clinicient or another EHR, and the exact record fields read and written are not documented for any of them.
Information is stored on servers in the United States, and the privacy policy states that neither Silna nor payers transfer or store it outside the country, while data moves to insurers and customer record systems as the service requires. Browser sessions that handle patient data run in isolated environments.
There is no choice of region and no option to run the platform in a customer's own cloud, and tenant boundaries across the database, the phone processors such as Twilio and Deepgram, and the model services that produce a result are not described.
The sales team quotes the platform on authorization volume and the modules a practice needs, not on seats, and a practice can buy prior authorizations, benefit checks and monitoring together or separately. The record sync is configured to the practice's own process with no separate public integration rate.
Volume bands, minimum commitments, module prices and an implementation fee are not listed, and the homepage's $35 copay and $12 evaluation figures are mockups of a patient quote rather than prices. A buyer can see which factors drive the price before calling, but not the rate to apply to its own volume.
The platform serves specialty provider groups, with applied behavior analysis, physical, speech and occupational therapy, behavioral health and psychiatry named among its fields. Customers include autism providers AnswersNow, Behavior One Autism Solutions and Autism Learning Partners, physical therapy groups Elite Alliance Physical Therapy and HealthPro Heritage, and Allergy & ENT Associates.
AnswersNow supplies the measured example in virtual autism therapy, and a physical therapy workflow covers functional outcome documents and remaining authorized visits. The company counts more than 1,000 payers across 50 states. Measured results exist for one specialty deployment, with none for physical therapy, ENT or individual payers.
Alternatives to Silna
Other products that do the same kind of work as Silna, and what each one does.
Cair Health sells AI agents for billing teams that manage insurance claims from eligibility through payment. Chloe checks notes, coverage information and payer requirements before submission.
Collectly sells AI revenue cycle automation for patient financial work across the visit. Before the visit, pre-visit intake automation handles digital check-in, insurance card capture, coverage checks, demographic validation and consents.
Claim Health sells an AI platform for home-based care agencies (home health, hospice, home care and other post-acute providers) that now presents itself as a care continuity platform running from referral to discharge.
Needletail AI verifies dental insurance eligibility and benefits for group practices and DSOs and writes the results into the practice management system before the patient arrives.
Mandolin automates the administrative work required to start and sustain specialty drug treatment. Its agents read referrals and clinical records, investigate insurance benefits, estimate patient costs, assemble prior authorization submissions and follow claims through payer portals and calls.
Smart Data Solutions runs claims, authorizations, appeals, enrollment and correspondence work for health plans, third party administrators and providers on what it calls an AI native platform, relaunched in July 2026, that joins intake, decision intelligence and workflow orchestration.
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 |
|---|---|---|---|---|
|
Contact for pricing
|
Quoted on authorization volume and the modules selected, not on seats. | No business associate agreement or eligible tier is stated in the terms or privacy policy. | No onboarding or integration fee is listed. Silna sets up the sync with the practice's EHR to follow its existing workflow. | Vendor Published |
Silna prices the platform by authorization volume and selected modules. Minimum volumes, module rates and onboarding charges are not listed.