AgencyFuse.ai Artificial Intelligence Policy

1. Purpose and Scope

This Artificial Intelligence Policy (the "AI Policy") governs the access to and use of all artificial intelligence features and capabilities made available within the AgencyFuse platform, including without limitation the AgencyFuse Assist conversational AI assistant, the Match AI-powered commission reconciliation engine, the Correct AI-driven data quality module, and any other feature that uses machine learning, large language models ("LLMs"), natural language processing, or automated inference to generate, classify, or act on data (collectively, "AI Features").

This AI Policy applies to all customers, users, administrators, and authorised representatives of Customer organisations who access the AI Features (collectively, "Users"). It supplements and is incorporated into the AgencyFuse End User Licence Agreement ("EULA"), Terms of Service ("ToS"), and Privacy Policy. In the event of conflict between this AI Policy and those documents, the more restrictive provision applies. Terms not defined herein carry the meanings assigned in the EULA.

2. Definitions

The following terms have the meanings set out below when used in this AI Policy:

1. "AI Features" means all artificial intelligence, machine learning, large language model, natural language processing, embedding, classification, or automated inference capabilities made available through the AgencyFuse platform, including AgencyFuse Assist, Match, and Correct.
2. "AI Output" means any text, data, recommendation, workflow trigger, action, summary, report, or other result generated by an AI Feature in response to an Input or automated process.
3. "AgencyFuse Assist" means the conversational AI assistant module within AgencyFuse that allows Users to query connected data sources using natural language and, where configured, to trigger automated actions or workflows.
4. "Input" means any data, query, prompt, instruction, document, or information submitted by a User or automated process to an AI Feature for processing.
5. "Customer Content" means all data, records, and information uploaded, transmitted, or made accessible to the AgencyFuse platform by or on behalf of a Customer, including AMS data, policyholder records, commission statements, and CRM data.
6. "Hallucination" means an AI Output that contains information that is factually incorrect, invented, or not supported by the underlying data provided to the AI Feature.
7. "Human Review" means the process by which a qualified human professional reviews, validates, and assumes responsibility for an AI Output before it is acted upon or communicated to a third party.
8. "Sub-processor" means any third-party entity engaged by Synatic to process Customer Content in connection with providing the AI Features, including third-party LLM providers.
9. "Synatic" means Synatic, Inc., the company that develops and operates the AgencyFuse platform. AgencyFuse is a product sold by Synatic, Inc., a corporation incorporated in the State of Georgia, USA, with its principal office at 44 Milton Avenue, Alpharetta, GA 30009.

3. Nature and Limitations of AI Features

1. AI Features Are Assistive Tools, Not Decision-Makers. The AI Features are designed to assist Users by surfacing data, identifying patterns, and generating outputs intended to inform human decision-making. AI Features do not constitute professional advice of any kind, including insurance advice, legal advice, compliance advice, financial advice, or actuarial advice. No AI Output should be treated as a final, authoritative, or legally binding determination without independent Human Review.
   IMPORTANT NOTICE: AI Outputs, including those generated by AgencyFuse Assist, may be incomplete, inaccurate, outdated, or otherwise unsuitable for the purpose for which they are used. Synatic expressly disclaims any warranty, representation, or guarantee as to the accuracy, completeness, fitness for purpose, or reliability of any AI Output. Users and Customers assume full responsibility for any action taken, or omitted, in reliance on an AI Output.
2. Known Limitations. Users must understand and account for the following inherent limitations of AI Features before relying on any AI Output:
   • AI models may produce Hallucinations — outputs that appear plausible but are factually incorrect or unsupported by source data.
   • AI Features operate on the data made available to them. Outputs are only as reliable as the quality, completeness, and timeliness of the underlying Customer Content. AgencyFuse's Correct module is designed to improve data quality but cannot guarantee absolute accuracy.
   • AI models are trained on data with a fixed knowledge cut-off date and may not reflect current regulations, policy terms, carrier guidelines, or market conditions.
   • AgencyFuse Assist, when configured to trigger actions or workflows, operates based on interpreted instructions. Ambiguous, incomplete, or unclear inputs may result in unintended actions.
   • AI Features may not have visibility into information that is not connected to or synchronised with the AgencyFuse platform.
3. Human-in-the-Loop Requirement. The AI Features within AgencyFuse are designed to augment and enhance the workflows of Users, not to replace their expertise, judgement, or decision-making authority. AI Features surface data, identify patterns, and generate outputs to inform and support Users. The responsibility for every professional decision, recommendation, or action taken in the course of agency operations remains at all times with the User and the Customer organisation.
   
   For any use case involving: (a) communications with insureds, prospective clients, or carriers; (b) binding, endorsing, or modifying coverage; (c) claims-related recommendations; (d) commission disbursements or financial reconciliations; or (e) any regulated or legally consequential act — Users must apply Human Review before relying on or acting upon an AI Output. The fact that an AI Output has been generated by AgencyFuse Assist or any other AI Feature does not, by itself, constitute approval, validation, or endorsement by Synatic.
   
   Users should treat AI Outputs as a starting point for informed decision-making, not as a substitute for professional analysis. Where an AI Output conflicts with a User's own knowledge, experience, or independent verification, the User's professional judgement should prevail.
4. Data Access — Read-Only Operations. The AI Features within AgencyFuse are designed to read, augment, and interpret data from connected source systems and Customer Content. AI Features do not write to, modify, delete, or otherwise alter records within any connected source database, AMS, CRM, or third-party system. All data interactions performed by AI Features are non-destructive and read-only with respect to the underlying source data.
   
   Where AgencyFuse Assist or other AI Features are configured to trigger automated actions or workflows that result in updates to connected systems, such actions are executed through defined integration pathways and only upon explicit User instruction or pre-configured automation rules authorised by the Customer. Any such write-back actions are distinct from the AI Features' core data processing functions and remain subject to the Human Review obligations set out in Section 3.3.

4. Acceptable Use of AI Features

1. Permitted Uses. Subject to the terms of the EULA, ToS, and this AI Policy, Users may use the AI Features for the following purposes in connection with their legitimate insurance agency operations:
   • Querying and summarising AMS and CRM data to support account management, renewal preparation, and producer workflows.
   • Identifying data quality issues, missing fields, duplicate records, and synchronisation exceptions in connected systems.
   • Processing and reconciling carrier commission statements using the Match AI module, subject to Human Review of flagged exceptions and confidence-scored mismatches.
   • Generating internal reports, data summaries, and analytics to support agency operations and leadership reporting.
   • Identifying policies, accounts, or clients that may require follow-up, subject to independent verification.
   • Automating routine, non-regulated data workflow tasks where the User has validated the configuration and assumes responsibility for outcomes.
2. Prohibited Uses. Users must not use the AI Features for any of the following purposes. These restrictions exist to protect policyholders, third parties, regulatory compliance, and the integrity of the AgencyFuse platform:
   • Generating or transmitting binding insurance quotations, policy binders, endorsements, cancellations, or coverage confirmations to clients or carriers without Human Review and appropriate authorised sign-off.
   • Making or presenting coverage recommendations to insureds as though they constitute professional insurance advice without disclosure of AI involvement and independent review.
   • Processing, storing, or transmitting Protected Health Information ("PHI") as defined under HIPAA, or Social Security Numbers, payment card data, or government-issued identity numbers, through the AI Features or as part of any Input.
   • Automating regulated claims-handling activities, including reserving recommendations, coverage determinations, or denial communications.
   • Attempting to reverse-engineer, extract, or access the underlying model weights, training data, prompt engineering, or system architecture of any AI Feature.
   • Using AI Features to engage in, facilitate, or conceal fraud, misrepresentation, or any activity that violates applicable insurance laws, regulations, or codes of conduct.
   • Circumventing, manipulating, or adversarially prompting AI Features to produce outputs that violate this AI Policy, the EULA, or applicable law.
   • Using AI Outputs as the sole basis for employment decisions, discriminatory underwriting, or any use case that violates applicable federal or state anti-discrimination law.

5. Data, Privacy, and Security

1. Customer Content and AI Processing. When a User submits an Input to an AI Feature, the underlying Customer Content included in that Input is processed by AgencyFuse's AI infrastructure, which may include third-party AI Sub-processors. Synatic processes Customer Content solely to provide, maintain, and improve the AI Features and the broader AgencyFuse Services, as set out in the EULA.
2. AI Sub-processors. Synatic may engage third-party LLM providers and AI infrastructure vendors as Sub-processors in connection with the AI Features. A current list of AI Sub-processors is available upon written request to [email protected]. Synatic will provide at least thirty (30) days' advance notice of any material changes to AI Sub-processor arrangements. Customers are responsible for ensuring that their use of the AI Features is consistent with their own privacy obligations and vendor management requirements under applicable law, including the GLBA Safeguards Rule and any applicable state insurance privacy regulations.
3. Model Training. Synatic does not use Customer Content, including Inputs submitted to AgencyFuse Assist or any other AI Feature, to train, fine-tune, or improve foundation models operated by third-party LLM providers, unless Customer has provided express written consent to such use. Synatic may use de-identified, aggregated, and anonymised data derived from platform usage to improve AgencyFuse product functionality, consistent with the rights granted in the EULA.
4. Sensitive Data Restrictions. Users must not submit the following categories of data as Inputs to any AI Feature:
   • Protected Health Information as defined under HIPAA.
   • Social Security Numbers, driver's licence numbers, passport numbers, or other government-issued identification numbers.
   • Payment card numbers, bank account numbers, or routing numbers.
   • Biometric data.
   • Any data whose processing in AI systems is prohibited by applicable regulation or Customer's own privacy policies.
   
   The EULA explicitly prohibits uploading data subject to HIPAA or similar regulatory requirements to the AgencyFuse platform. This restriction extends to all AI Feature Inputs.
5. Retention of Inputs and Outputs. Inputs submitted to AI Features and the corresponding AI Outputs may be retained by Synatic for a limited period for purposes including service delivery, quality assurance, security monitoring, and product improvement. Synatic's data retention practices are governed by the Privacy Policy. Users should avoid including unnecessary personal data in Inputs and should consult their organisation's data retention obligations when using AI Features in connection with regulated information.

6. AI in Insurance Operations — Special Considerations

1. Regulatory Compliance Remains the User's Responsibility. The use of AI tools in insurance operations is subject to evolving regulation at both the federal and state level. Without limitation, relevant frameworks may include the NAIC's Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, state-specific AI fairness and explainability requirements, New York Department of Financial Services Circular Letter No. 1 (2019) on AI use, and emerging state laws governing automated decision-making. Synatic does not warrant or represent that the AI Features, or any specific use of them, satisfies any such regulatory requirement. Customers are solely responsible for ensuring their use of the AI Features is compliant with all applicable laws and regulations.
   IMPORTANT NOTICE: AgencyFuse and its AI Features do not constitute a compliance or regulatory solution. No AI Output should be interpreted as legal, compliance, or regulatory guidance. Customers should seek independent legal and regulatory counsel with respect to their AI-related compliance obligations.
2. E&O Risk Awareness. Users using AgencyFuse Assist or other AI Features in connection with client-facing activities should be aware that reliance on AI Outputs without appropriate Human Review may expose the agency to errors and omissions liability. Synatic strongly recommends that Customer organisations review their E&O insurance coverage and internal procedures in light of their use of AI tools.
3. Agentic Actions — Heightened Review Obligations. Where AgencyFuse Assist or other AI Features are configured to trigger automated actions, including but not limited to updating AMS records, initiating workflows, or generating communications, Users must ensure that appropriate review and approval controls are in place before any such action takes effect in a live environment. Synatic shall not be liable for errors, omissions, or unintended consequences arising from automated actions taken by AI Features that have not been subject to appropriate Human Review and authorisation.

7. Disclaimers of Liability

1. No Warranty on AI Outputs. TO THE FULLEST EXTENT PERMITTED BY APPLICABLE LAW, SYNATIC EXPRESSLY DISCLAIMS ALL WARRANTIES, EXPRESS, IMPLIED, STATUTORY, OR OTHERWISE, WITH RESPECT TO AI FEATURES AND AI OUTPUTS, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, NON-INFRINGEMENT, OR UNINTERRUPTED AVAILABILITY.
2. No Professional Advice. AI Outputs generated by AgencyFuse AI Features do not constitute, and must not be relied upon as, professional insurance advice, legal advice, compliance advice, financial advice, actuarial advice, or any other form of regulated professional advice. Users who require such advice must consult a qualified professional.
3. Limitation of Liability. Synatic's aggregate liability arising out of or related to the AI Features, AI Outputs, or this AI Policy shall not exceed the amounts specified in the limitation of liability provision of the EULA. Synatic shall not be liable for any indirect, incidental, special, consequential, punitive, or exemplary damages arising out of or related to any User's reliance on an AI Output, including damages arising from regulatory enforcement actions, E&O claims, client complaints, or data errors.
4. Third-Party AI Models. The AI Features may be powered in whole or in part by third-party AI models and services. Synatic does not control the underlying behaviour, outputs, or model characteristics of such third-party systems and cannot guarantee consistency, availability, or accuracy of outputs generated by third-party models. Synatic's relationship with third-party AI providers is governed by separate agreements, and such providers are not parties to the Customer's agreement with Synatic.

8. User and Customer Responsibilities

By accessing or using AI Features, each User and Customer acknowledges and agrees to the following obligations:

• Review and validate: Implement adequate internal processes to review, validate, and, where required, seek Human Review of AI Outputs before acting upon them or relying on them for any regulated, client-facing, or legally consequential purpose.
• Data quality: Ensure that Customer Content submitted to AI Features is accurate, complete, and lawfully held. Poor data quality will degrade the quality of AI Outputs and remains the Customer's responsibility.
• Authorised use only: Use AI Features only for the purposes authorised under the EULA, ToS, and this AI Policy. Do not attempt to circumvent safety measures, usage restrictions, or content policies.
• Sensitive data: Do not submit sensitive personal data categories identified in Section 5.4 as Inputs to any AI Feature.
• Regulatory compliance: Independently comply with all applicable laws and regulations governing the use of AI tools in insurance operations, including training staff on regulatory obligations.
• Incident reporting: Promptly notify Synatic at [email protected] upon becoming aware of any AI Feature misuse, unexpected behaviour, data exposure, or regulatory inquiry related to AI Features.
• Staff training: Ensure that personnel using AI Features are appropriately trained on the limitations described in Section 3 of this AI Policy and on the Customer's own internal AI governance requirements.

9. Updates to This Policy

Synatic reserves the right to modify, update, or replace this AI Policy at any time to reflect changes in the AI Features, applicable law, regulatory guidance, or industry best practices. Material changes will be communicated to Customers via email notice to the registered account administrator or via in-product notification, with at least fifteen (15) days' prior notice before the amended policy takes effect. Continued use of AI Features after the effective date of any update constitutes acceptance of the revised AI Policy.

10. Contact and Reporting

For questions regarding this AI Policy, to request a list of current AI Sub-processors, or to report a concern relating to the AI Features, please contact:

• AI Policy and Privacy — [email protected]
• Security Incidents — [email protected]
• Mailing Address — Synatic, Inc., 44 Milton Avenue, Alpharetta, GA 30009
• Platform — agencyfuse.ai

Questions regarding AI-related regulatory compliance obligations specific to your agency should be directed to qualified legal counsel and are outside the scope of support Synatic is able to provide.