Silna Health vs SPRY (2026)

Silna works with more than 1,000 payers across 50 states, calling insurers for benefits and authorization status and submitting through their portals. On the provider side it connects to the EHR and practice management systems ABA providers already use, CentralReach and Catalyst among them, so insurance work runs alongside the clinic's existing setup. The list of practice systems is short and centered on ABA platforms, and whether authorization results flow back into them is unclear.

SourceSilna Health, silnahealth.com, /aba-therapy, engineering.silnahealth.com/posts/autonomous-browser-agents-enterprise and /posts/building-phone-agents-healthcareread 2026-10-09

Claims leave SPRY through a Claim.MD clearinghouse connection, with ERA files posted back automatically. Orders from ModMed arrive in SPRY and charges return to ModMed over HL7 once care is done, and Zocdoc bookings sync straight into the schedule. A Zapier connection reaches more than 5,000 other apps. Payments run through Stripe, Square, Paya, FortisPay and GoCardConnect, texting and reminders through Twilio and Emitrr, faxes through eFax, home exercise programs through Wibbi, and CRM through Mailchimp, Keap and the Zoho products.

Sourcespryhealth.ai/integrationsread 2026-10-09

The prior authorization lifecycle runs from expiration tracking and weekly reminders through form generation, submission and continual payer follow up. It branches as requests for additional information or resubmission raise alerts, denials move to appeal and Medicaid plan switches are flagged so authorizations can be resubmitted to the new plan. For ABA the same lifecycle covers initial assessments, concurrent reviews and reauthorizations. Behind it, document validation rules combine with AND and OR logic into configurations, and a graph decides which rules apply to each document type.

SourceSilna Health, silnahealth.com/prior-authorization, /insurance-monitoring, /aba-therapy and engineering.silnahealth.com/posts/document-validationread 2026-10-09

An authorization runs as a chain of steps. The eligibility check flags whether the plan needs prior authorization, the agent reads the therapist's note for diagnosis, goals, treatment plan, visit count and medical necessity, then completes and submits the payer workflow and tracks approved, used and remaining visits against the expiration date.

Anything it cannot finish goes to staff with the payer details, reference IDs, documentation and request history attached. One patient record carries the work across the scribe, scheduling, claim and denial steps, so a signed note feeds both the authorization and the claim.

Sourcespryhealth.ai/prior-authorizationread 2026-10-09

Work starts without staff: weekly expiration reminders, recurring eligibility and benefit checks, real time monitoring of every active patient's coverage, automatic detection of new insurance and Medicaid plan switches, and phone agents that call payers to check whether a submitted authorization was approved. For ABA, authorized hours remaining and upcoming reauthorization deadlines are tracked continuously so renewals go out before therapy is interrupted.

SourceSilna Health, silnahealth.com/insurance-monitoring, /prior-authorization, /aba-therapy and engineering.silnahealth.com/posts/building-phone-agents-healthcareread 2026-10-09

Work starts from events in the clinic day. Running eligibility routes a patient into the authorization workflow when the plan requires it, and benefits are checked daily so the front desk sees every scheduled patient's coverage each morning. A logged cancellation sends the open slot to eligible waitlist patients, who can accept after hours. Incoming faxes are read, classified and filed to the right chart and workflow. Outbound, webhooks tell connected systems when patient, appointment, case and SOAP note activity happens.

Sourcespryhealth.ai/schedulerread 2026-10-09

Silna maintains an always current database of each payer's prior authorization requirements per specialty, kept by its Payor Intelligence team from payer bulletins and the thousands of authorizations it processes each week, and applies it to form generation and document validation so providers never track the rules themselves. Predictive Document Intelligence reads diagnosis reports, treatment plans and progress notes with multimodal models and checks them against those criteria before submission, ABA treatment plans and progress reports included, and Silna reports a 98 percent first pass acceptance rate for requests it validates.

SourceSilna Health, silnahealth.com/prior-authorization, /aba-therapy, /resources/best-prior-authorization-software and engineering.silnahealth.com/posts/document-validationread 2026-10-09

The agents draw on the clinic's own record at each step. Before a visit the scribe brings forward prior notes, goals, baselines, range of motion history and plan of care context, and the authorization agent pulls what each payer asks for out of the therapist's note. Payer and clinic rules are applied to eligibility, authorization and claim scrubbing, and payer submissions run on workflows trained payer by payer. The agents work from the EMR record itself, rather than a separate index or knowledge base built over the clinic's documents, so their context moves with every signed note.

Sourcespryhealth.ai/agentic-scriberead 2026-10-09

What persists sits on Silna's side: the payer requirements database and the growing record of reviewer corrections that shapes later submissions. The agents work from those shared stores and each authorization's own record, with no memory of a clinic or patient that has its own scope or lifetime.

SourceSilna Health, silnahealth.com and engineering.silnahealth.com/posts/beyond-the-agentread 2026-10-06

The scribe learns how each clinician documents and applies those repeated preferences to future notes, and approved corrections, prior actions and outcomes make the next case easier. That learning is kept per clinician, with no set lifetime and no screen where a clinic can review or clear it. The patient history the scribe loads before each visit comes from the EMR's own notes and measures, so it is the clinical record rather than a separate memory.

Sourcespryhealth.ai/agentic-scriberead 2026-10-09

Staff stay in the loop where judgment matters. When document validation finds a gap against the payer's criteria, the provider reviews it and decides whether to update the documents or submit as they are, and requests for more information, resubmissions and appeals come back to the team as alerts. Behind the agents, Silna sends a set share of submissions to its own QA reviewers, who can override and correct the agent's work. Routine requests go out without staff touching each one, so there is no clinic sign off before a standard submission reaches the payer.

SourceSilna Health, silnahealth.com/prior-authorization, /resources/best-prior-authorization-software, engineering.silnahealth.com/posts/document-validation and /posts/beyond-the-agentread 2026-10-09

The Scheduling Agent books nothing until staff approve. It generates appointment sets from the plan of care, staff review the options and pick one, and that approval locks the plan before visits are created and confirmations go out. Clinics set the booking rules, provider availability, active hours and blocked time it works within, and can override its matches.

Clinicians review, edit and sign every scribe note, with flags for body part mismatches, conflicting measurements and missing details before sign off. On trained payer workflows the authorization agent submits supported requests itself, and anything outside them goes to staff with the case history attached.

Sourcespryhealth.ai/schedulerread 2026-10-09

Security starts with SOC 2 Type II certification through Vanta, operation under HIPAA and a pledge to treat patient confidentiality as a priority. On the agent side, each browser session is tied to a single provider through row level security, so one clinic's agent cannot reach another clinic's records. There is no single sign on or role management for a clinic's own staff.

SourceSilna Health, silnahealth.com and engineering.silnahealth.com/posts/autonomous-browser-agents-enterpriseread 2026-10-06

Access inside SPRY is role based, with clinical, billing and operational permissions kept separate, and data is encrypted at rest and in transit with audit logging. BAAs are available for covered entities. API and agent access is approved per partner, scoped per tenant and held to role based permissions and read and write boundaries. The EHR is ONC certified health IT with a CHPL listing. Alongside that, SPRY's security rests on HIPAA alignment rather than a SOC 2, HITRUST or ISO 27001 attestation, and there is no single sign on or SCIM provisioning for clinic users.

Sourcespryhealth.ai/trust-and-complianceread 2026-10-09

Clinics follow each authorization from submission to approval on a dashboard, with approvals passed along as they arrive. Inside Silna, every inference call and tool execution of a browser agent is saved as a pipeline step and each session gets a quality score from 0 to 100, a record that serves Silna's own engineers and reviewers. Clinics see status and outcomes rather than the agent's step by step trail.

SourceSilna Health, silnahealth.com/prior-authorization and engineering.silnahealth.com/posts/beyond-the-agentread 2026-10-06

Authorization reporting follows approval status, payer turnaround, denials, expirations and follow up needs across locations. Each case that reaches staff carries its request history, payer reference IDs and documentation, so a biller picking up an exception sees what was already tried. Audit logging is part of the platform's security controls, though clinics work from these reports and case histories rather than a searchable log of every agent action.

Sourcespryhealth.ai/prior-authorizationread 2026-10-09

The service is hosted. Its browser agents operate in sandboxed sessions on Silna's own self hosted Kasm Workspaces deployment, which isolates each session from the next and keeps agent activity off the clinic's own machines. There is no region choice, customer cloud or on premises option.

SourceSilna Health, silnahealth.com and engineering.silnahealth.com/posts/autonomous-browser-agents-enterpriseread 2026-10-06

SPRY runs as a hosted service on its own infrastructure, with no region choice, single tenant or customer cloud option and no on premises install. When a clinic leaves, billing stops without penalties or lock in fees, and one designated user keeps seven years of read only access to its records at no cost.

Sourcespryhealth.ai/trust-and-complianceread 2026-10-09

Silna sells one Care Readiness Platform built from prior authorizations, benefit checks, insurance monitoring and payor discovery, with prior authorizations priced at $35 each. These work as parts of one service rather than separate agents a clinic switches on, and there is no template or agent library. Customers include Behavior One Autism Solutions, Autism Learning Partners, ABS Kids, AnswersNow, HealthPro Heritage and Elite Alliance Physical Therapy.

SourceSilna Health, silnahealth.com and /resources/best-prior-authorization-softwareread 2026-10-09

SPRY ships named agents for separate jobs: Agentic Scribe for documentation, the Prior Authorization Agent, the Scheduling Agent for plan of care booking and cancellation backfill, and AI Fax for reading and filing incoming documents, with Claim, AR and Compliance Intelligence on the billing side. Each does its own work on the shared record, and the Scheduling Agent also runs in connected scheduling workflows outside the full platform. AI Scribe comes with the Plus plan.

Sourcespryhealth.ai/spry-airead 2026-10-09

Models are picked for each payer portal after testing cost, context window, intelligence and speed, and production traffic can be split between models that perform equally, for example half to Haiku 4.5 and half to GPT 5.4. The phone agents use Deepgram for speech recognition and ElevenLabs for speech, and the customer has no choice of model.

SourceSilna Health, engineering.silnahealth.com/posts/beyond-the-agent and /posts/building-phone-agents-healthcareread 2026-10-06

The models behind the scribe and the agents are undisclosed, and clinics have no choice of model. SPRY puts the weight on context instead: patient history, payer rules and the schedule each agent draws on when it acts.

Sourcespryhealth.ai/spry-airead 2026-10-09

Authorizations can be followed from submission to approval through an API as well as Silna's dashboard, which gives a clinic's other systems a way to track them. There is no public API reference, SDK or MCP server, and the agents themselves work through payer portals, payer phone lines and Silna's own app.

SourceSilna Health, silnahealth.com/resources/best-prior-authorization-softwareread 2026-10-09

A public developer site documents a REST API, v1 current and v2 in beta, across scheduling (clinics, providers, availability, appointments, cancellation reasons, waitlists), patient and clinical records (patients, cases, documents, insurance, referring providers, SOAP notes) and revenue operations (eligibility, prior authorization, staff, tasks).

Webhooks report patient, appointment, case and SOAP note activity, and a quickstart walks through creating an appointment. Credentials and a sandbox go to approved customers and partners on request. SPRY MCP sits over these APIs as a governed layer for outside agents, with no public endpoint or tool list.

Sourceapi-docs.spryhealth.careread 2026-10-09

An in house Experiments platform sets what share of prior authorizations each portal agent submits with no human touch and routes a share of agent submissions to human QA, comparing human and AI results, and specialist reviewers flag minor or major errors to build a dataset of where the agent was wrong. Models are tested per portal in a sandbox on production infrastructure for duration, token use and accuracy, and every session gets a score from 0 to 100 for execution quality, apart from its final outcome.

SourceSilna Health, engineering.silnahealth.com/posts/beyond-the-agentread 2026-10-06

Approved corrections, prior actions and outcomes feed back into the agents, and the scribe checks each draft for inconsistencies before sign off. There is no evaluation harness or test suite a clinic can use to measure agent accuracy before or after a change, and the sandbox SPRY offers is for partners testing API integrations.

Sourcespryhealth.ai/spry-airead 2026-10-09

Browser agents work payer portals on their own, logging in with 2FA, navigating complex legacy interfaces, entering clinical data and recording approvals, across Blue Cross, Aetna, United and dozens of other portals, inside sandboxed browser sessions. They now handle nearly half of all prior authorizations and are trained to stop and ask for help rather than guess when stuck. Phone agents navigate payer phone menus to retrieve plan details.

SourceSilna Health, engineering.silnahealth.com/posts/autonomous-browser-agents-enterprise, /posts/beyond-the-agent and /posts/building-phone-agents-healthcareread 2026-10-06

For supported payers the authorization agent completes the payer's form, attaches the documentation and submits the request itself, on trained workflows for Carelon and BCBS, UnitedHealthcare and Humana, with more payer portals on the way. A supported request takes about 90 seconds against twenty minutes by hand. Whether that runs through a browser session on the payer portal or a direct payer connection is unclear.

Sourcespryhealth.ai/prior-authorizationread 2026-10-09