Friendly
Also known as: Friendly Cares, Friendly Inc., Omniscient
AI infrastructure layer for insurance that turns complex medical and financial documents into auditable, source linked decision insights for underwriting and claims.
Friendly is an enterprise AI platform for life and health insurers, reinsurers and brokers that turns a full case file, medical records, handwritten doctor's notes, lab results, X-rays, EHR extracts, APS reports, financial documents and legal papers, into a structured, decision ready case at up to 7,000 pages an hour.
Imported documents are classified, grouped and cleaned, the medical data is extracted and connected to SNOMED, ICD-10 and LOINC codes, and the result is presented as a case dashboard with a HealthGrid of notable impairments, a case timeline of medical events, a narrative summary, a discrepancy and pattern recognition view that flags mismatches between medical evidence and disclosures, and a chat assistant that answers questions by searching across every document in the case, with each item linked back to its source page.
Solutions cover underwriting, claims across life, death, disability, long term care and workers' compensation, and reinsurance treaties, where the Omniscient assistant searches and filters across a treaty collection. The platform auto adjudicates claims and supports straight through processing of new business, with assessors and underwriters brought in to review a case or approve a payment, and it produces a customizable summary report a team can export into its own systems. It is multilingual, offers a PII anonymizer, and runs either in Friendly's hosted infrastructure or in the customer's own private cloud, with HIPAA, GDPR and SOC 2 compliance stated.
Friendly is based in San Francisco, founded and led by Natasha Alexeeva, built with insurers and reinsurers across North America, Europe, Africa and Asia, and sold by demo with no published pricing. It does not publish an API, model choice, or workflow orchestration tooling.
Vendor details
Canonical URL
https://www.friendlycares.com
Category
Enterprise operations agent
Funding status
Independent, purpose built for insurance and financial services. Specific funding figures were not retrieved this session.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Positioned as a software only AI infrastructure layer that ingests complex medical data, financial documents, and handwritten notes and fits into insurance decisioning workflows. Specific integration connectors were not enumerated on the retrieved pages.
Sources & related URLs
Research sources
Agentic Index coverage score
5.0 / 14 capabilities · 36%
| Integrations & Tool Calling | Partial |
|---|---|
|
The about page states the platform digitizes any document from any source, and the features page says Friendly imports and processes the customer's medical data and lets a team share a case as a customizable summary report to feed their own system or print out. Ingestion of case files and export of a report is the whole documented integration surface: no connector, catalog, custom tool route or authenticated action in an outside system is named. No agent action in an outside system is documented, and the published surface is thin, file based ingestion and export with no system named. SourceFriendly, friendlycares.com/features and friendlycares.com/aboutread 2026-09-07 |
|
| Workflow Orchestration | Partial |
|
The about page states the platform digitizes documents, produces summary reports, auto adjudicates claims and enables new business straight through processing, with people called on to review a case or approve a payment when needed; the underwriting page shows the processing pipeline as classify, group, cleanup and structured report. That is a fixed multi step pipeline the vendor runs end to end on a case, not orchestration the customer composes: no workflow builder, no multi agent execution, no named runtime and no multi participant workflow is documented. SourceFriendly, friendlycares.com/about and friendlycares.com/underwritingread 2026-09-07 |
|
| Knowledge Grounding & RAG | Full |
|
The features page documents that imported case documents are OCR extracted, structured into cards grouped by diagnosis, abnormal labs, vitals over time and physical exams, connected to SNOMED, ICD-10 and LOINC codes, and presented in a case dashboard where every item links to its source; an AI chat assistant then answers questions by searching across all documents in a case, and a treaties advanced query searches and filters across a collection of reinsurance treaties. The structured, coded extraction persists as the case record and stays queryable by the assistant, which is a maintained retrieval structure over the customer's documents rather than context assembled per run. SourceFriendly, friendlycares.com/featuresread 2026-09-07 |
|
| Human Oversight & Guardrails | Partial |
|
Claims are auto adjudicated, and the about page says people are called on when needed to review a case or approve a payment; the underwriting and features pages present recommendations on next best actions for a human decision. That is a one sentence claim that an approval step exists for payments and exceptions. No review surface, approval queue, configuration of what routes to a human, or the conditions that trigger review is documented. SourceFriendly, friendlycares.com/about and friendlycares.com/underwritingread 2026-09-07 |
|
| Security, Identity & Governance | Partial |
|
The features page says the platform adheres to HIPAA, GDPR and SOC 2 guidelines, and the underwriting page says deployment ensures HIPAA, GDPR and SOC 2 compliance; neither asserts an audit or certification, so the posture is alignment rather than attestation. The features page also names advanced encryption, multi layered access controls, continuous threat monitoring and a PII anonymizer, but no identity integration (SSO, SAML, SCIM) and no named access model (roles, RBAC, permissions) is published. There is no trust or security subdomain and no product documentation, and the site is a Wix marketing site with a privacy statement. SourceFriendly, friendlycares.com/features and friendlycares.com/underwritingread 2026-09-07 |
|
| Observability & Auditability | Partial |
|
Every piece of information in the case dashboard links to its original source, so a reviewer can click through to where it came from, and a case timeline follows each medical event. That is provenance for extracted outputs, showing what the extraction says and where it came from, not a runtime record of what the agent did: no log of agent actions, decisions, model calls or adjudication steps, no reconstruction of why an auto adjudication landed where it did, and no audit export is documented. SourceFriendly, friendlycares.com/featuresread 2026-09-07 |
|
| Memory & State Persistence | Not documented |
|
The persistent case history and timeline belong to the case itself. The case dashboard, timeline and structured extraction are the customer's case record, the application's data model, and deleting them would delete the case. No memory layer with a stated scope and lifetime, per user, per team or per workflow, is documented, and no session state for the chat assistant is published. SourceFriendly, friendlycares.com/featuresread 2026-09-07 |
|
| Deployment & Data Residency | Full |
|
The features page states the platform runs whether hosted in Friendly's secure infrastructure or within the customer's own private cloud, and the underwriting page says it easily deploys on private cloud infrastructure. Customers can deploy into their own private cloud or use vendor hosting, so a buyer with a different requirement gets a different answer. No hosting region for the vendor hosted option, no deployment documentation and no selection surface is published, and the option rests on a marketing sentence. SourceFriendly, friendlycares.com/features and friendlycares.com/underwritingread 2026-09-07 |
|
| Prebuilt Agents / Templates / Packs | Not documented |
|
The site presents three solution lines, underwriting, claims and treaties, and a set of named features on one platform: HealthGrid and Focus Grid, case dashboard, chat assistant, medical summary, narrative summary, case timeline, pattern recognition and PII anonymizer. These are modules and views of one product rather than separate agents a buyer selects and adopts as they stand, and no template library or catalog of prebuilt agents is published. The Omniscient treaty assistant is a named product on the press page, but it is one product, not a pack. SourceFriendly, friendlycares.com/features and friendlycares.com/newsread 2026-09-07 |
|
| Triggers & Channel Coverage | Partial |
|
Processing starts when a case file is imported: the features page says work begins once Friendly imports the customer's medical data, and the about page says documents are digitized from any source. Case arrival is the only documented trigger; no event, schedule, webhook or inbound queue is documented, and the chat assistant is invoked by a user. SourceFriendly, friendlycares.com/features and friendlycares.com/aboutread 2026-09-07 |
|
| Model Flexibility & Routing | Not documented |
|
The features page says Friendly's model is trained to recognize medical data and can be trained on additional languages, describing the vendor's own models. No model provider is named, and no customer or admin choice of model or routing is documented. SourceFriendly, friendlycares.com/featuresread 2026-09-07 |
|
| APIs / SDKs / MCP Extensibility | Not documented |
|
No API, SDK, MCP server or other interface for calling the platform from outside is documented, and there is no developer documentation. The only outbound surface is a summary report a team can export to feed its own system, which is a file handed back, not an interface for driving Friendly. The vendor calls the product a software only infrastructure layer, which suggests embeddability, but nothing documents it. SourceFriendly, friendlycares.com/featuresread 2026-09-07 |
|
| Testing, Debugging & Optimization | Not documented |
|
The features and underwriting pages document pattern recognition that uncovers mismatches between medical data and disclosures, discrepancy highlighting and fraud signals. The artifact under test there is the customer's case, data quality on the applicant's file, not the agent, and no evaluation harness, scored test cases, release gate, controlled loop or debugging trace for the platform's own outputs is published. SourceFriendly, friendlycares.com/features and friendlycares.com/underwritingread 2026-09-07 |
|
| Browser / Computer-use | Not documented |
|
The platform OCR extracts and structures uploaded documents, handwritten notes, lab results, X-rays and EHR exports, at up to 7,000 pages an hour, and presents them in its own dashboard; that is ingestion into the case record, not computer control. No browser, desktop or computer control driven by the agent at run time is documented. SourceFriendly, friendlycares.com/features and friendlycares.com/underwritingread 2026-09-07 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Contact for pricing
Cost watchouts
As an infrastructure layer over sensitive medical and financial data, deployment likely requires an integration and compliance engagement. Cost is scoped per deployment and volume.
Variable cost rationale
Likely scoped by document or case volume across underwriting and claims; no public rate card, so exposure is moderate and not precisely determinable.
Sales call required
Yes, required for paid access
Free / trial
No public free tier; enterprise engagement
Lowest paid plan
Not public
Key ambiguities
No public pricing. Whether billing is per document, per case, per seat, or platform tier is not disclosed, and documentation is thin.
Related vendors
- 11th Estate — Agentic platform whose AI engine scans markets, matches an…
- Adopt AI — AI CPA firm whose agents run reconciliations, close, AP and AR, and…
- Aera Technology — Decision intelligence platform where an always on agent executes…
- Akro AI — On premise operational intelligence that automates document heavy…
- alfred_ — AI executive assistant that triages the inbox overnight and scores…
- Alloy.ai — Commerce intelligence system for consumer brands that unifies four…
Alternatives to Friendly
The closest documented capability profiles to Friendly among enterprise operations agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Akro AI6.5 / 14Adds documented Memory & State Persistence
- Balerion AI4.5 / 14Fuller documented coverage on Workflow Orchestration
- effie.ai4.5 / 14Adds documented Prebuilt Agents, Templates & Packs
- Porters5.5 / 14Adds documented Prebuilt Agents, Templates & Packs
- Smart Bricks3.5 / 14A lighter documented profile than Friendly
- Kenmei7.0 / 14Adds documented Prebuilt Agents, Templates & Packs
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded