Nym

Nym assigns codes to qualifying patient encounters and can send them directly into billing after configuration and user-acceptance testing. That puts the burden on the buyer's controls: coding quality, coverage, exception routing, denials, and ongoing audits need to hold on real records before straight-through billing is enabled. [S1][S2][S3]

Best for: A health system or physician group with a high-volume in-scope service line, reliable historical data, a coding lead, a technical owner, and the capacity to run a controlled UAT and ongoing audit program.

Not for: A buyer seeking general clinical summarization, a light overlay with no EHR/billing integration work, a team without accountable HIM or revenue-cycle governance, or an organization that cannot keep a human route for exceptions, audits, denials, and code corrections.

Configure coding for a service line

• Trigger: The provider chooses an in-scope specialty, facility, and coding workflow.
• Inputs: Historical charts, associated coding data, local or payer-specific guidelines, SOPs, chart formats, and buyer requirements. [S2]
• Output: A configured engine and production-flow design intended to match the buyer's coding practices. [S2]
• Human checkpoint: The coding lead reviews the manually coded discovery sample and gives feedback before configuration proceeds. [S2]
• Pilot measure: Agreement with the buyer's validated coding sample, exception/unsupported-record rate, critical-error rate, and readiness of the escalation design.

Code qualifying encounters for billing

• Trigger: A daily discharge record or other covered patient encounter enters the configured integration flow.
• Inputs: The patient record and configured coding rules/ontologies. [S2][S3]
• Output: A coded encounter with supporting clinical findings, guideline references, and code-assignment rationale; Nym says qualifying records route to billing. [S1][S3]
• Human checkpoint: The revenue-cycle owner monitors sampled results, coding coverage, denials, and downstream billing effects; records outside the approved route need a named human disposition.
• Pilot measure: Coding coverage, turnaround time, audit accuracy, coding-correction rate, denial rate, and time from discharge to bill.

Audit and expand safely

• Trigger: The buyer receives UAT or live performance data and considers go-live or another specialty/facility.
• Inputs: A random audit sample, Nym-coded records, buyer coding results, accuracy findings, critical errors, compliance findings, and technical integration test results. [S2]
• Output: A go-live, remediation, hold, or scoped-expansion decision.
• Human checkpoint: The coding lead and technical project owner validate the audit method and findings; the accountable revenue-cycle leader accepts or rejects expansion.
• Pilot measure: Accuracy by code family and encounter type, missed material issues, exception turnaround, correction latency, audit-trail completeness, and post-billing rework.

1. The provider selects a narrow specialty and facility, names an executive sponsor, coding lead, technical project manager, and integration team, and collects representative historical charts and coding data. [S2]
2. Nym documents a discovery phase that reviews buyer coding practices and designs the production flow; its team manually codes a set of charts for buyer feedback before engine configuration. [S2]
3. The engine is configured to the buyer's requirements and connected to the designed technical workflow. [S2]
4. In UAT, Nym says the buyer sends daily discharge records, Nym audits a random subset of its coded output, and the parties test the integration. [S2]
5. Only after the buyer accepts the agreed accuracy, compliance, turnaround, exceptions, and billing controls should qualifying encounters take the direct route; the provider continues sampling and monitoring live performance.

The decisive trade-off is autonomy. Straight-through coding can reduce manual handling for covered encounters, but it raises the importance of upstream configuration, UAT, audit design, exception routing, access controls, and the ability to stop or correct a bad flow before it affects billing.

Verified facts

• Nym documents an autonomous coding engine, six named specialties/service lines, audit-ready documentation for each code, and an existing-system integration position. [S1]
• Nym documents an implementation process that uses customer guidelines, SOPs, historical charts, associated coding data, discovery feedback, configuration, integration, UAT, and technical testing. [S2]
• The implementation page states a 3–6 month typical timeline and describes stated go-live criteria of 95%+ UAT accuracy and 12-hour turnaround time. [S2]
• Nym describes CLU as machine learning plus rules-based clinical ontologies and says its audit trail shows clinical findings, coding-guideline references, and code-assignment reasoning. [S3]

Vendor claims

• Nym states that its engine will autonomously code 50%–70% of complete records after go-live, with coverage varying by specialty, and that it maintains or exceeds a 95% accuracy threshold. Treat these as vendor statements to validate on the buyer's mix and audit method. [S2]
• Nym says it is HIPAA compliant and SOC 2 Type II certified/mapped to HITRUST. Obtain the current report, scope, period, exceptions, BAA, and terms; public marketing copy is not a security assessment. [S1]
• In Nym's published Inova story, Inova reports $1.3M lower annual ED coding costs, a 50% reduction in weekly ED DNFB, and more than 10% higher average revenue per ED encounter. Those are vendor-published customer results, not independent predictions or proof for another provider. [S4]

B2Bagents assessment

Nym’s public implementation detail gives a more testable operating model than a generic “AI coding” claim: buyer-provided historical records and guidelines, configured integration, audit sampling, named go-live conditions, and ongoing performance discussions. The exact risk is equally clear: Nym says qualifying encounters go directly to billing. A strong buyer pilot therefore treats coding accuracy, coverage, denial impact, exceptions, and reversibility as one control system—not as separate technology and HIM projects. [S1][S2][S3]

Material unknowns

• Public pricing, minimum term, trial terms, scope unit, implementation fees, change-request fees, and termination terms.
• Exact support for the buyer's specialties, facilities, encounter types, code families, EHR/billing interfaces, payer rules, and routing paths.
• Data retention/deletion, data residency, encryption key ownership, subprocessors, model providers, support access, audit-log export, incident commitments, and BAA language for production clinical records.
• The precise accuracy calculation, audit sample, coverage denominator, exception thresholds, correction workflow, and performance remedies that will govern the buyer’s contract.
• Independent evidence that Nym will improve the buyer’s coding cost, revenue capture, DNFB, or denial rate; B2Bagents did not perform a hands-on test.

Good-fit conditions: The buyer has a narrow initial service line, sufficient volume, reliable EHR and billing data, local coding rules that can be documented, an accountable coding lead, a technical owner, and authority to run parallel audit before direct-to-billing activation.

Poor-fit conditions: The scope is undefined, coding practices vary without documented rules, upstream charts are incomplete, the buyer lacks a safe human path for exceptions, or the team cannot examine payer denials and audit results after activation.

Predictable failure modes and controls:

• Local guidance or chart conventions can be misrepresented in configuration. Use representative historical charts, document exceptions, and require buyer feedback on discovery samples and UAT results. [S2]
• A good overall accuracy percentage can conceal a material error cluster. Segment audits by facility, specialty, encounter type, code family, payer, and severity; define stop conditions before launch.
• Routing directly to billing can propagate a bad interface or status mapping. Test hold, correction, resubmission, duplicate handling, downtime, and reconciliation paths with the EHR and billing owners. [S1][S2]
• A case study can hide scope differences. Treat Inova's published outcomes as diligence leads, then baseline and measure the buyer's own cost, coding, DNFB, and denial data. [S4]

Assign a revenue-cycle executive as accountable sponsor, a coding/HIM lead as quality owner, and a technical project manager as integration owner. Begin with one facility and service line. Establish a frozen baseline for volumes, code distribution, manual workload, audit accuracy, correction rate, DNFB, denials, days to bill, and downstream rework.

Nym’s process calls for customer coding data and historical charts, buyer-specific configuration, production-flow design, integration, UAT, and go-live. Its page names FHIR and major EHRs, but the public material does not prove the buyer's exact connector, field mapping, write scope, billing-system interface, or failure recovery. [S2]

Before expanding, demonstrate record reconciliation across each system, error queues, code correction and rebill paths, audit-log retrieval, data export, access removal, downtime operation, and the buyer’s ability to pause routing without losing records.

Nym’s product page includes vendor statements of HIPAA compliance and SOC 2 Type II certification/mapping to HITRUST. Its public privacy policy documents website and contact-data practices, service-provider sharing, safeguards, retention principles, and rights; it does not supply the complete clinical-data terms a covered entity needs. [S1][S5]

Obtain the current SOC report and bridge letter, BAA, DPA, subprocessor and model-provider list, security architecture, test summary, identity and access controls, encryption details, logging/export controls, retention/deletion choices, residency, support-access policy, and incident/SLA commitments. In the operating design, separately authorize system integration, configuration changes, exception disposition, billing release, and audit review.

No public pricing was located in the first-party sources reviewed. Request a quote that identifies whether charges depend on facility, specialty, encounter volume, coding coverage, implementation, integration, support, additional service lines, or other usage. Also request minimum commitment, renewal increases, overage treatment, availability commitments, performance remedies, data export, and termination assistance.

Price the pilot as a controlled specialty-and-facility deployment. A lower per-encounter price can be outweighed by configuration, interface, audit, security-review, and parallel-run work that the buyer must still own.

Run a parallel UAT for one facility and service line before direct billing. Use a stratified, buyer-approved sample that includes common records plus incomplete documentation, conflicting notes, unusual procedures, new provider templates, corrected charts, high-value encounters, and known historical coding disputes.

For each record, capture the final human-reviewed code, Nym’s result, stated rationale, source documentation, confidence or routing decision if available, reviewer disposition, materiality, correction time, and downstream billing/denial outcome. Track coding coverage, accuracy by segment, critical-error rate, false direct-route rate, exception rate, turnaround time, reviewer minutes, DNFB, denial rate, and days to bill.

Set written thresholds and stop conditions before beginning. Stop, hold, or narrow the flow if a material category cannot be audited, an unsupported record reaches billing, an interface reconciliation fails, a critical error pattern exceeds the agreed rate, required security evidence is unavailable, or the buyer cannot reproduce the rationale and correction history for an encounter.

Compare Nym with alternatives by coding scope and control model first: specialty coverage, direct-to-billing versus coder validation, customer-specific configuration, audit evidence, EHR/billing integration, exception handling, reporting, security commitments, pricing unit, and buyer-run quality results. A comparable product should solve coding operations—not merely draft clinical text or summarize documentation.

1. Which encounters are eligible for direct billing in our first scope, and what routes every other record? Ask for the exact coverage definition, confidence/routing rules, exception owner, and no-record-left-behind reconciliation. [S1][S2]
2. How will accuracy be measured for our specialties and code families? Agree on the sample, comparator, severity taxonomy, denominator, audit independence, correction process, and contractual thresholds before UAT. [S2]
3. What exactly changes in our EHR and billing systems? Inspect data fields, FHIR/interface mappings, permissions, status changes, error handling, downtime behavior, audit records, and rollback/hold controls. [S2]
4. What security and privacy commitments cover our production records? Obtain current SOC/BAA/DPA evidence and confirm retention, access, model use, subprocessors, support access, export, deletion, and incident response. [S1][S5]
5. What would make us stop or expand the pilot? Write evidence thresholds for coverage, accuracy, critical errors, turnaround, correction effort, DNFB, denials, and audit-trail completeness before the first data transfer.