AI & Clinical Decision Support Disclaimer

Synexar, Inc. ("Synexar," "we," "us," or "our") operates Synexar, an AI-native clinical-documentation platform for gastrointestinal (GI) endoscopy and anesthesia. This Disclaimer explains the intended use — and the limits — of the artificial-intelligence ("AI") features in Synexar (the "Service").

This Disclaimer is published for the benefit of the licensed clinicians, ambulatory surgery centers, physician practices, hospitals, and other healthcare organizations (each a "Facility") that use the Service, and for the patients those Facilities serve. It should be read together with our Privacy Policy (https://synexar.ai/legal/privacy), Acceptable Use Policy (https://synexar.ai/legal/aup), Subprocessor List (https://synexar.ai/legal/subprocessors), and AI Policy (https://synexar.ai/legal/ai-policy).

1. What Synexar Is

Synexar is a clinical-documentation platform that helps licensed clinicians produce, structure, and sign procedure reports and related workflow artifacts (case lists, charge tickets, quality-registry submissions). The Service accepts inputs from the clinician — free-text dictation, voice transcription, structured form entry, and integrated device feeds — and assembles those inputs into a draft document for the clinician's review, edit, and signature.

The clinician — not Synexar — is the author of the report. AI output that has not been reviewed and signed by a clinician is a draft only.

2. What Synexar Is Not — Not a Medical Device

Synexar is not a medical device, a clinical decision support system in the regulated sense, a diagnostic tool, or a treatment-recommendation engine. The Service does not:

• Diagnose disease, recommend treatment, or evaluate the appropriateness of a clinical action;
• Acquire, interpret, or analyze medical images or physiologic signals for diagnostic purposes;
• Replace, supplement, or independently exercise clinical judgment;
• Substitute for direct clinician examination of the patient or the patient's record; or
• Provide guidance intended for use in time-critical or emergency situations.

Synexar has assessed the Service's clinical-documentation AI features against the criteria for the Clinical Decision Support ("CDS") software exemption from medical-device regulation under section 520(o)(1)(E) of the Federal Food, Drug, and Cosmetic Act, as interpreted by the FDA's final guidance, Clinical Decision Support Software, issued January 6, 2026 (re-issued January 29, 2026), which revised the FDA's prior September 2022 final guidance.

Based on that assessment, Synexar believes the Service satisfies all four prongs of the exemption because it:

1. Is not intended to acquire, process, or analyze a medical image, or a signal from an in-vitro diagnostic device or a pattern/signal-acquisition system. The Service operates on clinician-entered structured data and clinician dictation — not on raw device images or physiologic signals.
2. Is intended to display, analyze, or print medical information normally communicated between healthcare professionals — procedure findings, narrative text, and candidate codes.
3. Is intended to provide recommendations, not specific directives — the clinician chooses whether to accept, edit, or reject every AI suggestion. (Indeed, the Service does not provide treatment recommendations at all.)
4. Is intended to enable the clinician to independently review the basis for each output so that the clinician does not rely primarily on the software to make a clinical decision.

The clinician can review the basis for every Synexar-generated section of every report by inspecting (i) the source dictation, (ii) the structured inputs, and (iii) the change log retained for the case in the audit trail.

This is Synexar's good-faith regulatory assessment of the Service as currently designed. It is not a determination by the FDA and is not legal or regulatory advice to any Facility. Each Facility remains responsible for its own regulatory compliance.

3. Synexar's Role: Business Associate, Not a Covered Entity

Synexar is a Business Associate as that term is defined at 45 C.F.R. § 160.103, acting on behalf of the Facilities that license the Service. Synexar is not a Covered Entity: it does not provide healthcare, does not submit claims to payers, and does not maintain the medical record as the entity of record.

For clarity, using the definitions at 45 C.F.R. § 160.103:

• A Covered Entity is a health plan, a healthcare clearinghouse, or a healthcare provider that transmits health information electronically in connection with a covered transaction. The Facility and its clinicians are the Covered Entities in the Synexar relationship.
• A Business Associate is a person or entity that creates, receives, maintains, or transmits protected health information ("PHI") to perform a function or activity on behalf of a Covered Entity. Synexar is the Business Associate.

Synexar's handling of PHI on behalf of each Facility is governed by a Business Associate Agreement ("BAA") executed with that Facility, consistent with the HIPAA Privacy, Security, and Breach Notification Rules.

4. Clinician Responsibility

The clinician of record accepts clinical responsibility for the final signed report. Before any AI-generated content becomes part of the medical record, the clinician is responsible for:

• The clinical accuracy of every statement in the final signed report;
• The medical decisions reflected in the report (diagnosis, intervention, plan, follow-up);
• Compliance with applicable standards of care, payer requirements, and the clinician's own credentialing and licensure obligations;
• Verification of patient identity, procedure performed, and findings; and
• Selection and appropriateness of billing codes (CPT, ICD-10, modifiers) included in the report.

A report is not part of the medical record until a clinician has reviewed it, corrected it as needed, and approved it through electronic signature.

The disclaimer language that renders on report PDFs and in the pre-sign signature confirmation is maintained in Synexar's Report Disclaimer Block (a PDF-surface reference document) and is not reproduced here.

5. AI-Generated Content — Known Limitations

The Service uses large language models (currently Azure OpenAI GPT and Anthropic Claude, with provider failover) to expand clinician dictation into draft narrative sections and to extract structured findings from free text. AI-generated content carries known limitations, including:

• Hallucination — AI may generate text that is plausible but factually unsupported by the clinician's input. The clinician must verify every clinical fact before signature.
• Omission — AI may fail to incorporate every relevant detail from the clinician's input. The clinician must confirm completeness.
• Misattribution — AI may incorrectly assign findings to anatomical regions, laterality, or temporal sequence. The clinician must verify anatomical and temporal accuracy.
• Categorization inference — AI may suggest categorizations (e.g., polyp morphology, withdrawal time) that do not match the clinician's intent. The clinician must verify.
• Provider drift — AI model behavior changes over time as providers update their models; Synexar's deterministic billing rules and text-normalization layer mitigate but do not eliminate this risk.

No AI-generated text in any Synexar report is final until the clinician signs the report. Signing the report attests that the clinician has reviewed and accepts every statement.

6. Billing-Code Suggestions Are Administrative, Not Clinical

Billing-code suggestions — candidate CPT codes and modifiers — are administrative and financial outputs, not clinical decision support. They are produced by a deterministic rules engine that applies National Correct Coding Initiative ("NCCI") edits and bundling logic to clinician-confirmed structured findings. No large language model selects, decides, or finalizes a billing code. The LLM's role is limited to producing reviewable clinical narrative and structured findings; the code-selection path is a separate, deterministic, auditable subsystem.

Candidate codes are presented for review by the clinician or the Facility's billing professional. Final code selection is the responsibility of the provider and the Facility's billing team. Synexar does not submit claims and is not responsible for the accuracy of any claim a Facility submits.

7. Not For Emergency or Critical Use

Synexar is documentation software for use after a procedure or clinical event has occurred. It must not be used as the primary or sole source of information in any time-critical clinical situation — including acute decompensation, intra-procedural complication, or any circumstance in which delay could result in patient harm. In such situations, clinicians must rely on direct examination, validated medical devices, and established institutional protocols.

8. Voice Transcription

The Service uses Azure Whisper (Azure Speech) to transcribe clinician dictation. Transcription accuracy depends on audio quality, ambient noise, accent, and clinical vocabulary. The clinician must review the transcribed text before it is included in any signed report. Synexar does not warrant transcription accuracy and is not liable for the consequences of unreviewed transcription errors.

9. AI Providers & Data Handling

The Service uses contracted third-party AI providers operating in the United States, each under a Business Associate Agreement or equivalent data-protection terms:

• Azure OpenAI (United States; configured for zero data retention) — narrative generation and structured-data extraction;
• Anthropic Claude (United States; BAA executed March 15, 2026) — narrative generation; and
• Azure Whisper / Azure Speech (United States) — speech-to-text transcription of clinician dictation.

All AI processing occurs within the United States. PHI is not used to train any third-party provider's foundation models. The current, authoritative list of AI subprocessors is published in our Subprocessor List (https://synexar.ai/legal/subprocessors). Our governance practices for these providers are described in our AI Policy (https://synexar.ai/legal/ai-policy).

PHI is handled as described in Section 3 and is governed by the executed BAA between Synexar and the Facility. Patients with questions about how their PHI is used should contact their Facility directly. See also our Privacy Policy (https://synexar.ai/legal/privacy).

10. No Patient-Facing Recommendations

Synexar is licensed for use by credentialed clinicians and authorized Facility staff only. No Synexar-generated output is intended for direct patient consumption as medical guidance. Patient-facing materials (such as post-procedure instructions) generated using the Service must be reviewed and approved by the clinician before delivery to the patient.

11. State-Specific Notices

Synexar offers the Service to Facilities in the United States only. The following notices reflect Synexar's understanding of state laws relevant to AI-assisted clinical documentation. Where a state law imposes an AI-use disclosure duty on the provider or Facility (rather than on the upstream software vendor), Synexar supplies disclosure templates the Facility may adopt; the disclosure obligation itself runs to the Facility.

• Texas — Under the Texas Responsible Artificial Intelligence Governance Act (HB 149) and SB 1188, the duty to disclose AI use to patients runs to the deploying Facility or provider, not to Synexar as the upstream vendor. Synexar supplies disclosure templates Facilities may adopt. SB 1188's requirement that EHR data be stored in the United States is satisfied by Synexar's US-only data residency.
• California — Under AB 3030, which governs generative-AI patient clinical communications, the Service's workflow places a licensed clinician in the review-and-sign loop for every report. Synexar therefore relies on AB 3030's clinician-review exemption. (Synexar has no California sites at launch.)
• Colorado — The Colorado AI Act originally enacted as SB 24-205 was repealed and replaced by SB 26-189, signed May 14, 2026 and effective January 1, 2027. SB 26-189 is a notice-only statute and does not impose the earlier algorithmic-discrimination impact-assessment regime. Synexar has no Colorado Facilities at launch, so this notice is informational.

This list is not exhaustive and does not constitute legal advice. Each Facility remains responsible for compliance with the laws applicable to its own jurisdiction and practice.

12. Governance Frameworks

Synexar designs and monitors the Service's AI features with reference to recognized healthcare-AI governance frameworks, including the AMA Augmented Intelligence Principles (physician-in-the-loop), the Coalition for Health AI (CHAI) governance playbooks (8 playbooks, published May 2026), and the NIST AI Risk Management Framework 1.0 together with NIST AI 600-1 (the Generative AI Profile). Our full governance approach is described in our AI Policy (https://synexar.ai/legal/ai-policy).

13. Reporting Issues

A clinician who suspects a Synexar report contains a clinical error must (a) correct the report before signing, and (b) report the issue to Synexar at [email protected] so the underlying pattern can be investigated. Synexar maintains an audit log of all AI-generated content, all clinician edits, and the final signed version of every report for the retention period required by applicable law and the BAA.

14. Changes to This Disclaimer

Synexar may update this Disclaimer to reflect changes in the Service, applicable law, or guidance. Material changes will be notified to Facility primary contacts at least 30 days before taking effect, and will be reflected by an updated effective date at the top of this page. The most current version is published at https://synexar.ai/legal/ai-disclaimer.

15. Contact

Questions about this Disclaimer or the Service's AI features may be directed to Synexar's HIPAA Privacy & Security Officer:

Raghuram Vadapally — Co-Founder, Principal Architect; HIPAA Privacy & Security Officer
Synexar, Inc.
5900 Balcones Drive, STE 100, Austin, TX 78731
[email protected]

Synexar's Chief Executive Officer is Dr. Ashutosh Gupta (CEO, CMO & Chief Business Strategist).

This Disclaimer is a publication candidate pending counsel sign-off (Olender Feldman). It is not legal advice.