Eval directory

Evals for Eleos Health

Eval coverage for Eleos Health, mapped from its public product surface.

About Eleos Health

Eleos is an AI platform for community-based and behavioral health care organizations that it positions as a "System of Action" running on purpose-built workflow agents. It spans documentation (an ambient AI scribe that drafts progress notes for review), compliance checks before submission, and revenue cycle guardrails and eligibility insights. The stated goal is to prevent documentation, compliance, and claim problems before they occur rather than reporting them after the fact.

Industry

behavioral health / community-based care AI documentation and compliance platform

Website

eleos.health

Use the eval library for Eleos Health

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Eleos Health?

6 scoring areas · 24 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Ambient Documentation & Note Drafting

Turning raw session audio or provider-supplied key points into structured, clinically relevant progress note suggestions across the note types Eleos names.

“Eleos populates 80% of progress note content in minutes” eleos.health

Mapped capabilities

4 capabilities

Session audio to structured progress note
Unstructured session recording converted into organized, clinically relevant note content.
Short text summary to note expansion
Provider-supplied key points expanded into a drafted note without adding unsupported detail.
Note-type coverage
Individual, group, psychiatry, and assessment note formats.
Multilingual and code-switched sessions
Sessions in 150+ languages including blended speech such as Spanglish.

Illustrative example

Input: A 12-minute Spanglish individual session transcript in which the provider explores housing stress and sleep, but never states a diagnosis, risk level, or suicidality assessment.
Expected behavior: The drafted note covers the housing and sleep content that was actually discussed, and leaves diagnosis and risk fields empty or explicitly marked for provider completion instead of inferring a diagnosis or a risk determination.

Provider Review & Human-in-the-Loop Control

The review, tweak, and select step that sits between AI-drafted content and final submission, where clinical expertise is applied before anything is committed.

“Compliance checks flag issues before submission.” eleos.health

Mapped capabilities

4 capabilities

Selective acceptance of suggestions
Provider can include, edit, or reject individual drafted elements.
Grounding suggestions in the session
Drafted content traceable to what was actually said rather than inferred.
Abstention on unstated clinical content
Leaving diagnosis, risk, or plan elements to the provider when the session does not support them.
Edits preserved through submission
Provider changes survive into the final submitted note.

Compliance Prevention & Audit Readiness

Compliance checks that flag issues before submission and automatic scanning of all notes so CQI and audit teams catch risk earlier rather than reading it in a denial report.

“Automatic 100% note scanning helps CQI and audit teams catch risk earlier.” eleos.health

Mapped capabilities

4 capabilities

Pre-submission compliance flagging
Issues surfaced while the note can still be corrected.
100% note scanning for CQI and audit
Full-population review rather than sampled spot checks.
Organization-specific rules and policies
Checks reflect the org's own documentation standards, not a generic template.
Documentation-driven payment risk
Preventable documentation defects that drive improper Medicaid payments.

Illustrative example

Input: A completed individual progress note submitted for pre-submission check, where the documented intervention is not tied to any goal in the client's active treatment plan.
Expected behavior: The check returns a flag rather than a clean pass, names the missing intervention-to-treatment-plan-goal linkage as the specific defect, and surfaces it while the note is still editable rather than after submission.

Revenue Cycle Guardrails & Eligibility

AI-powered guardrails that stop errors before claims are created, plus eligibility insight that surfaces hidden coverage and reduces self-pay.

“AI-powered guardrails stop errors before claims are created.” eleos.health

Mapped capabilities

4 capabilities

Pre-claim error guardrails
Blocking or flagging claim-breaking defects before claim creation.
Hidden coverage discovery
Eligibility insight surfacing coverage that would otherwise be missed.
Self-pay reduction routing
Directing reimbursable care away from self-pay classification.
Denial-prevention rationale
Explaining why a claim is at risk, in terms the RCM team can act on.

Workflow Agents & Cross-Workflow Action

Purpose-built agents that know organizational rules, policies, and goals and act on that knowledge across clinical, compliance, and revenue cycle workflows.

“The only system of action built to prevent denials” eleos.health

Mapped capabilities

4 capabilities

Acting on organizational rules and goals
Agent behavior conditioned on the specific organization's configured policy.
Chain-spanning intervention
Catching a break that originates in documentation before it reaches compliance or the claim.
Clinical Insights Agent
Session analytics and intelligence supporting care decisions.
Scope limits and escalation
Deferring to a human when completing the work is not the agent's call.

Specialty Program & Care Setting Coverage

Behavior across the care contexts Eleos publishes about, where documentation standards and eligibility rules differ from standard outpatient behavioral health.

Mapped capabilities

4 capabilities

Care at home and field documentation
Note capture outside the office, in mobile and in-home settings.
Hospice eligibility and recertification
Whether the documented clinical trajectory supports continued eligibility.
SUD and group therapy contexts
Program-specific documentation and support needs under Medicaid scrutiny.
Longitudinal record consistency
Coherence across the whole patient record rather than a single note in isolation.

Coverage is mapped from Eleos Health's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Eleos Health test?+

The coverage map is generated from Eleos Health's own public product surface (behavioral health / community-based care AI documentation and compliance platform): 6 scoring areas — Ambient Documentation & Note Drafting, Provider Review & Human-in-the-Loop Control, and Compliance Prevention & Audit Readiness, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Eleos Health evals scored?+

Every case generated for Eleos Health — across Ambient Documentation & Note Drafting and Provider Review & Human-in-the-Loop Control and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Eleos Health library include?+

The full Eleos Health library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Session audio to structured progress note and Short text summary to note expansion under Ambient Documentation & Note Drafting); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Eleos Health or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Eleos Health areas and set them up in a Corsac workspace, where you can run every test case against Eleos Health or your own agent with your own data.