SCHEDULE B

ENTERPRISE SERVICE LEVEL AGREEMENT

Enterprise AI SaaS Platform Version 1.0 · Effective Date: 4 August 2026

This Enterprise Service Level Agreement (this "SLA") forms an integral part of the Terms of Service (the "Agreement") accepted electronically by the Customer (including by clickwrap acceptance during account registration or online checkout). This document is published at a permanent URL on ONLYAI’s website and is incorporated into the Agreement by reference. It applies between Dostart Ventures Limited ("ONLYAI", "Company", or "Provider") and the Customer identified in the applicable electronic Order Form ("Customer").

Capitalized terms not otherwise defined herein shall have the meanings assigned to them in the Agreement.

1. PURPOSE AND OBJECTIVES

1.1 Purpose

The purpose of this SLA is to define the operational service commitments applicable to the Services provided by ONLYAI.

This SLA establishes:

• minimum operational service levels;
• platform availability commitments;
• support obligations;
• incident response procedures;
• maintenance procedures;
• AI operational commitments;
• customer support standards;
• business continuity commitments;
• disaster recovery objectives;
• service reporting obligations;
• escalation procedures.

The Parties acknowledge that this SLA establishes commercially reasonable operational standards for the provision of enterprise AI software services and shall not be construed as guaranteeing uninterrupted availability of the Services or error-free operation.

1.2 Scope

This SLA applies solely to the production environment of the Services identified in the applicable Order Form.

Unless otherwise expressly agreed in writing, this SLA does not apply to:

• Beta Services;
• Preview Features;
• Alpha Releases;
• Proof-of-Concept environments;
• Sandbox environments;
• Customer-developed integrations;
• Third-Party Applications;
• Experimental AI models;
• Professional Services;
• Consulting Services.

1.3 Service Philosophy

ONLYAI provides an enterprise-grade software platform designed to assist Customer in automating and optimizing digital communications through AI-assisted technologies.

The Services are provided on a commercially reasonable efforts basis and are designed to deliver high availability, operational resilience and continuous improvement while recognizing the inherent characteristics of cloud computing environments, artificial intelligence technologies and third-party infrastructure providers.

Customer acknowledges that AI-assisted software differs fundamentally from deterministic software and therefore certain operational characteristics, including model behavior, inference latency and generated outputs, may vary over time.

2. INTERPRETATION

This SLA shall be interpreted consistently with the Agreement.

In the event of conflict, the order of precedence set out in Section 3 of the Agreement shall apply. This SLA shall prevail only in relation to service levels, availability, support, Incident management and Service Credits expressly governed by this SLA.

3. DEFINITIONS

For purposes of this SLA the following definitions shall apply.

"Availability"

means the percentage of time during a calendar month during which the Production Environment is capable of accepting authenticated customer requests and performing the core functionality of the subscribed Services.

"Business Continuity Event"

means an event materially affecting Provider's ability to provide the Services, including:

• infrastructure failures;
• regional cloud outages;
• cyber security incidents;
• natural disasters;
• force majeure events;
• denial of service attacks;
• failures of critical third-party suppliers.

"Critical Functionality"

means functionality necessary for Customer's ordinary business operations, including:

• authentication;
• AI message generation;
• dashboard access;
• workflow execution;
• notification processing;
• API processing (where applicable).

"Downtime"

means the period during which Critical Functionality is unavailable to Customer, excluding all Excluded Downtime identified in this SLA.

"Emergency Maintenance"

means maintenance reasonably necessary to:

• preserve platform integrity;
• remediate vulnerabilities;
• mitigate cyber threats;
• restore degraded services;
• comply with legal obligations;
• prevent material operational risks.

"Incident"

means any unplanned interruption, degradation or reduction in quality affecting the Services.

"Production Environment"

means ONLYAI's live operational environment made available to Customer under an active Subscription.

“Response Time”

means the period between:

(i) ONLYAI’s receipt of a valid Incident notification through an authorized support channel containing sufficient information reasonably necessary to identify and investigate the reported Incident; and

(ii) ONLYAI’s initial acknowledgement of the Incident and commencement of commercially reasonable triage or investigation activities.

Response Time does not constitute a guaranteed resolution, remediation, recovery or restoration time.

"Resolution"

means implementation of a permanent correction or a commercially reasonable workaround restoring affected functionality.

"Scheduled Maintenance"

means planned maintenance communicated to Customer in advance.

"Service Credit"

means a monetary credit calculated in accordance with Section 14.

Service Credits represent Customer's exclusive financial remedy for failure to satisfy the Availability Commitment.

"Supported AI Model"

means any artificial intelligence model integrated into the Services and designated by ONLYAI as supported for production use.

"Supported Cloud Provider"

means any cloud infrastructure provider designated by ONLYAI for hosting production workloads.

“Third-Party Dependency”

means any third-party infrastructure, software, platform, application programming interface, database, data storage service, artificial intelligence service, model provider, model routing service, cloud service, hosting service, content delivery network, domain name service, telecommunications service, authentication service, queueing service, caching service, monitoring service, communications service, integration service or other external technology or service not owned and directly controlled by ONLYAI upon which any functionality of the Services depends, including, where applicable, Supported Platforms and Subprocessors.

4. SERVICE MODEL

4.1 Enterprise SaaS Model

Customer acknowledges that the Services are delivered as a multi-tenant enterprise Software-as-a-Service platform.

ONLYAI is responsible for:

• operating the production environment;
• maintaining platform security;
• maintaining software infrastructure;
• monitoring platform health;
• maintaining APIs developed, owned and directly controlled by ONLYAI, excluding third-party APIs, Supported Platform APIs, external integration interfaces and other APIs not owned and directly controlled by ONLYAI;
• maintaining AI orchestration services;
• applying software updates;
• maintaining backup procedures;
• operating customer support.

Customer remains responsible for:

• internet connectivity;
• local networks;
• endpoint devices;
• browser compatibility;
• internal user administration;
• Creator onboarding;
• AI configuration;
• workflow configuration;
• Supported Platform credentials.

4.2 Shared Responsibility Model

The Parties acknowledge that operation of the Services follows a shared responsibility model.

ONLYAI shall remain responsible for the security and availability of the cloud-based software platform under its control.

Customer shall remain responsible for all operational decisions relating to the configuration and use of the Services, including the management of user accounts, AI automation settings, communication workflows, retention periods, access rights and compliance with Applicable Laws and Supported Platform requirements.

Nothing in this SLA transfers Customer's operational, regulatory or business responsibilities to ONLYAI.

4.3 Continuous Service Improvement

Customer acknowledges that ONLYAI continuously develops and improves the Services.

Accordingly, ONLYAI may implement:

• software enhancements;
• infrastructure optimizations;
• security improvements;
• AI model upgrades;
• workflow optimizations;
• new platform capabilities;
• performance improvements;
• monitoring enhancements;

provided that such changes do not materially diminish the overall functionality of the subscribed Services.

Routine improvements implemented pursuant to this Section shall not constitute a breach of the Agreement or this SLA.

5. SERVICE AVAILABILITY COMMITMENT

5.1 Availability Objective

Subject to the exclusions, dependencies, limitations and measurement methodology set forth in this SLA, ONLYAI shall use commercially reasonable efforts to maintain a Monthly Service Availability of not less than ninety-nine point five percent (99.5%) for the core Production Environment directly operated and controlled by ONLYAI.

The Availability Commitment applies solely to the availability of core production functionality within ONLYAI’s reasonable technical control and does not extend to the availability, performance, latency, continuity or interoperability of any Third-Party Dependency.

The Parties expressly acknowledge and agree that the Availability Commitment constitutes a service level objective and not an absolute representation, warranty or guarantee of uninterrupted, continuous or error-free operation.

For the avoidance of doubt, any unavailability, degradation, latency, interruption or loss of functionality caused directly or indirectly by a Third-Party Dependency shall be excluded from the calculation of Monthly Service Availability to the extent that the relevant event is outside ONLYAI’s reasonable control.

5.2 Availability Calculation

Monthly Service Availability shall be calculated as follows:

Availability (%) = ((Total Minutes in the Measurement Period − Downtime) ÷ Total Minutes in the Measurement Period) × 100

For purposes of this calculation, Downtime shall exclude any Excluded Downtime expressly identified in this SLA.

5.3 Measurement Period

Availability shall be measured on a calendar month basis unless otherwise agreed in the applicable Order Form.

5.4 Service Monitoring

ONLYAI shall maintain commercially reasonable monitoring systems designed to continuously assess the operational status, availability and performance of the Production Environment.

Operational monitoring may include, as appropriate:

• infrastructure health monitoring;
• application performance monitoring;
• API availability monitoring;
• AI service health monitoring;
• database monitoring;
• queue monitoring;
• security event monitoring; and
• automated alerting for critical service degradation.

Monitoring data maintained by ONLYAI shall constitute the authoritative source for determining compliance with the Availability Commitment, unless the Parties mutually agree that a manifest error has occurred.

6. ENTERPRISE SUPPORT SERVICES

6.1 Enterprise Support Commitment

Throughout the applicable Subscription Term, ONLYAI shall maintain an enterprise-grade customer support function designed to facilitate the continuous operation of the Services and assist Customer in resolving technical issues affecting the Production Environment.

Support Services are intended to address technical incidents relating to the operation of the Services and do not include legal advice, regulatory advice, business consulting, workflow design, Creator management, prompt engineering services (unless separately purchased), or other professional services.

Unless otherwise specified in the applicable Order Form, Support Services shall be provided in the English language.

6.2 Support Channels

Customer may submit support requests through one or more of the following channels designated by ONLYAI:

• Customer Support Portal;
• Designated Support Email;
• Enterprise Support Portal;
• Emergency Contact Channel (Enterprise Customers only);
• Customer Success Manager (where applicable).

ONLYAI may modify available support channels upon reasonable prior notice, provided that at least one electronic support channel remains continuously available.

6.3 Authorized Support Contacts

Customer shall designate one or more Authorized Support Contacts who shall serve as the primary liaison between the Parties regarding operational matters.

ONLYAI shall be entitled to rely upon instructions and requests received from Authorized Support Contacts.

Customer shall promptly notify ONLYAI of any changes to its Authorized Support Contacts.

6.4 Support Hours

Unless otherwise specified in an applicable Order Form:

Standard Support

Monday through Friday

09:00 – 18:00 UTC

excluding official public holidays observed in the Republic of Cyprus.

Emergency Support

Severity 1 – Critical Incident

A Critical Incident means an Incident resulting in the complete or substantially complete unavailability of Critical Functionality for a material proportion of Customer’s Authorized Users, where no commercially reasonable workaround is available.

Target Initial Response Time: within three (3) hours.

For the avoidance of doubt, the Target Initial Response Time constitutes a commitment to acknowledge the Incident and commence commercially reasonable triage and investigation activities. It does not constitute a guaranteed resolution, remediation, recovery or restoration time.

The applicable Response Time shall commence only upon ONLYAI’s receipt of a valid Incident notification containing sufficient information reasonably necessary to identify and investigate the reported Incident. Any period during which ONLYAI is awaiting information, access, logs, credentials, cooperation or other reasonably necessary assistance from Customer shall be excluded from calculation of the applicable Response Time.

7. INCIDENT MANAGEMENT

7.1 General Principles

ONLYAI shall maintain documented Incident Management procedures designed to ensure timely identification, classification, escalation, investigation, mitigation and resolution of Incidents affecting the Services.

Incident Management activities shall be performed using commercially reasonable practices consistent with modern cloud service operations.

7.2 Incident Lifecycle

Each Incident shall generally progress through the following lifecycle:

• Detection;
• Registration;
• Initial Classification;
• Technical Investigation;
• Containment (where applicable);
• Escalation;
• Resolution or Workaround;
• Verification;
• Closure;
• Post-Incident Review (for Major Incidents).

ONLYAI reserves the right to adjust Incident handling procedures where reasonably necessary to respond to evolving operational circumstances.

7.3 Incident Prioritization

Incidents shall be prioritized based upon:

• operational impact;
• number of affected users;
• availability of workarounds;
• security implications;
• business continuity impact;
• regulatory implications;
• likelihood of data loss.

ONLYAI shall determine the initial Incident severity in good faith and may revise the severity classification during investigation where appropriate.

8. INCIDENT SEVERITY CLASSIFICATION

Severity 1 – Critical Incident

A Critical Incident exists where one or more of the following occurs:

• complete loss of production availability;
• inability of all Authorized Users to access the Services;
• complete failure of AI message generation;
• widespread failure of workflow execution;
• material corruption of Customer Data;
• confirmed cybersecurity incident affecting the Production Environment;
• confirmed Personal Data Breach requiring immediate containment;
• complete failure of authentication services.

Target Initial Response

Within three (3) hours.

Target Continuous Updates

At commercially reasonable intervals having regard to the nature, severity and duration of the Incident, and in any event following any material change in Incident status, identification of a material root cause, implementation of a material workaround or restoration of the affected functionality.

Resolution Objective

Continuous commercially reasonable efforts until restoration of Critical Functionality.

Severity 2 – High Incident

A High Incident materially degrades production functionality but does not render the Services entirely unavailable.

Examples include:

• failure of a major software module;
• significant AI processing degradation;
• dashboard inaccessibility for a substantial portion of users;
• API failures materially affecting integrations;
• persistent performance degradation.

Target Initial Response

Within 2 Business Hours.

Target Resolution Objective

Within 1 Business Day, where reasonably practicable.

Severity 3 – Medium Incident

Examples include:

• degradation of non-critical functionality;
• reporting inaccuracies;
• delayed notifications;
• isolated integration failures;
• non-critical workflow failures.

Target Initial Response

Within 8 Business Hours.

Resolution Objective

Within 5 Business Days, subject to complexity.

Severity 4 – Low Incident

Examples include:

• cosmetic interface issues;
• documentation corrections;
• enhancement requests;
• usability improvements;
• minor defects with no material operational impact.

Target Initial Response

Within 2 Business Days.

Resolution shall occur in a future software release or within a commercially reasonable timeframe.

9. MAJOR INCIDENT MANAGEMENT

9.1 Major Incidents

A Major Incident means any Incident reasonably expected to:

• materially interrupt Customer's business operations;
• affect multiple enterprise customers simultaneously;
• involve significant cybersecurity risks;
• require executive-level coordination;
• result in material reputational risk;
• trigger mandatory regulatory notification obligations.

9.2 Major Incident Team

ONLYAI shall maintain internal procedures for the management of Major Incidents.

Where appropriate, a Major Incident response team may include representatives from:

• Engineering;
• Platform Operations;
• Information Security;
• AI Operations;
• Customer Success;
• Executive Management;
• Legal & Compliance.

9.3 Customer Communication

For Major Incidents, ONLYAI shall use commercially reasonable efforts to provide Customer with timely updates regarding:

• current operational status;
• mitigation activities;
• estimated restoration progress;
• available workarounds;
• anticipated next steps.

Nothing in this Section obligates ONLYAI to disclose Confidential Information, security-sensitive information, or details that could reasonably increase cybersecurity risks.

10. ESCALATION PROCEDURES

10.1 Operational Escalation

Where Customer reasonably believes an Incident is not receiving appropriate attention, Customer may request escalation through the following levels:

Level 1 — Support Engineer

↓

Level 2 — Senior Support Engineer

↓

Level 3 — Engineering Manager

↓

Level 4 — Head of Engineering / Platform Operations

↓

Level 5 — Executive Management (where appropriate).

Escalation shall not alter the applicable Incident Severity unless reclassified by ONLYAI.

10.2 Customer Cooperation

Customer shall cooperate with ONLYAI during Incident investigation by:

• providing accurate technical information;
• supplying logs where reasonably available;
• testing proposed workarounds;
• promptly responding to requests for clarification.

Failure to provide reasonably requested cooperation may affect ONLYAI's ability to meet the response or resolution objectives set forth in this SLA.

10.3 Root Cause Analysis

Following a Severity 1 Incident or other Major Incident, Customer may request a commercially reasonable Root Cause Analysis ("RCA").

Where an RCA is provided, it may include, as appropriate:

• a high-level description of the Incident;
• the identified root cause (if determined);
• corrective actions implemented;
• preventive measures intended to reduce the likelihood of recurrence.

The Parties acknowledge that RCAs are intended to facilitate operational transparency and continuous improvement and shall not constitute an admission of fault or liability by ONLYAI.

11. AI OPERATIONS

11.1 AI Service Model

Customer acknowledges that the Services utilize one or more artificial intelligence models supplied by ONLYAI and, where applicable, independent third-party AI providers.

The Services have been architected as an AI-assisted communication platform incorporating multiple interconnected software components, including AI inference services, orchestration engines, workflow engines, safety systems, moderation layers, analytics modules and cloud infrastructure.

Accordingly, temporary degradation affecting one individual component shall not automatically constitute a failure of the Services where the overall functionality of the subscribed Services remains substantially available.

11.2 AI Service Objective

ONLYAI shall use commercially reasonable efforts to maintain continuous availability of AI-assisted functionality throughout the Subscription Term.

The Parties acknowledge that AI-generated outputs are inherently probabilistic and that variations in model behavior, inference quality, response style or generation sequence do not, in themselves, constitute a Service Incident.

ONLYAI shall continuously monitor the operational health of AI-assisted services and implement commercially reasonable measures designed to maintain service continuity.

11.3 AI Inference Availability

Subject to the exclusions set forth in this SLA, ONLYAI shall use commercially reasonable efforts to ensure that AI inference services remain available for production workloads.

Where the primary AI inference provider experiences service degradation, ONLYAI may, at its sole discretion, implement one or more contingency measures, including:

• routing requests to an alternative supported AI model;
• temporary degradation of non-essential AI functionality;
• prioritization of production workloads;
• temporary limitation of optional AI features;
• implementation of queue management mechanisms.

Customer acknowledges that such contingency measures are intended to preserve overall platform availability and shall not constitute a breach of this SLA.

11.4 AI Safety Systems

ONLYAI maintains AI safety mechanisms intended to reduce the likelihood of unlawful, harmful or platform-non-compliant AI outputs.

Such mechanisms may include:

• prompt safety filters;
• prohibited content detection;
• age-related safety mechanisms;
• platform compliance rules;
• output moderation;
• escalation logic;
• confidence thresholds;
• anomaly detection.

ONLYAI reserves the right to modify or enhance such safety mechanisms at any time where reasonably necessary to improve platform security, regulatory compliance or operational integrity.

11.5 AI Model Lifecycle

ONLYAI continuously evaluates and improves the AI models and orchestration technologies utilized by the Services.

Accordingly, ONLYAI may from time to time:

• replace supported AI models;
• upgrade model versions;
• modify prompt engineering techniques;
• optimize routing logic;
• improve retrieval mechanisms;
• introduce additional safety controls;
• modify inference parameters.

Such modifications shall not require Customer approval provided that they do not materially reduce the subscribed functionality.

12. THIRD-PARTY DEPENDENCIES

12.1 Third-Party Infrastructure and Technology Dependencies

Customer acknowledges that the Services depend upon Third-Party Dependencies operated by independent providers and not owned or directly controlled by ONLYAI.

Such dependencies may include, without limitation:

(a) cloud and hosting infrastructure;

(b) external and third-party APIs;

(c) Supported Platform connectivity and integration services;

(d) managed database infrastructure;

(e) object storage and media storage services;

(f) content delivery networks;

(g) domain name and network security services;

(h) artificial intelligence model routing, inference and processing services;

(i) queueing, caching and asynchronous task processing infrastructure;

(j) telecommunications and internet service providers;

(k) authentication and identity services;

(l) monitoring, analytics and operational tooling; and

(m) other external infrastructure necessary for delivery of particular functionality.

ONLYAI shall not be responsible for any unavailability, degradation, latency, interruption, restriction or loss of functionality caused directly or indirectly by the failure, outage, degradation, modification, restriction, suspension, discontinuation or other unavailability of a Third-Party Dependency, to the extent that the relevant event is outside ONLYAI’s reasonable control.

ONLYAI shall use commercially reasonable efforts to mitigate material disruption and restore affected functionality where technically and commercially practicable. Nothing in this Section shall require ONLYAI to guarantee the performance of a Third-Party Dependency, maintain redundant providers for every dependency or immediately migrate to an alternative provider.

12.2 Third-Party AI and Model Routing Services

Customer acknowledges that ONLYAI may utilize independent third-party artificial intelligence infrastructure providers, model routing services, inference providers, foundation model providers and large language model providers in connection with the delivery of AI-assisted functionality.

AI requests may be transmitted through a third-party model routing or orchestration provider that selects, routes or otherwise facilitates access to one or more underlying artificial intelligence models. Accordingly, ONLYAI may not maintain a direct technical or contractual integration with every underlying model provider made available through such routing infrastructure.

Third-party AI providers and model routing services may modify:

(a) available models;

(b) routing logic;

(c) APIs;

(d) inference behaviour;

(e) response latency;

(f) rate limits;

(g) supported features;

(h) geographic processing arrangements; or

(i) service availability.

ONLYAI shall use commercially reasonable efforts to adapt the Services to material changes affecting such dependencies but does not warrant uninterrupted availability of any particular third-party AI provider, model routing service, foundation model or underlying AI model.

12.3 Supported Platforms

Customer further acknowledges that operation of the Services may depend upon Supported Platforms operated by independent third parties.

ONLYAI shall not be liable for any interruption, degradation or functional limitation resulting from:

• changes to Supported Platform APIs;
• changes to authentication mechanisms;
• changes to platform policies;
• restrictions imposed by Supported Platforms;
• suspension of Customer accounts by Supported Platforms.

12.4 Third-Party Dependency Incidents

Where an Incident affecting the Services is caused or materially contributed to by a Third-Party Dependency, ONLYAI shall use commercially reasonable efforts to:

(a) identify the affected Third-Party Dependency where reasonably practicable;

(b) assess the material impact of the Incident upon the Services;

(c) implement available mitigation measures to the extent technically and operationally within ONLYAI’s reasonable control;

(d) communicate material updates to Customer where the Incident has a significant and continuing impact upon Critical Functionality; and

(e) restore affected functionality following restoration of the relevant Third-Party Dependency, where further action by ONLYAI is reasonably required.

For the avoidance of doubt, ONLYAI shall not be responsible for resolving an Incident within systems, infrastructure, APIs, databases, platforms or services owned, operated or controlled by a third party and shall not be bound by any resolution, remediation, restoration or recovery time in respect of such Third-Party Dependency.

Any unavailability, degradation, latency, interruption or failure attributable to a Third-Party Dependency shall be treated in accordance with Section 17 (Excluded Downtime).

13. CHANGE MANAGEMENT AND RELEASE MANAGEMENT

13.1 Continuous Delivery

ONLYAI utilizes a continuous software development and deployment methodology designed to improve the Services throughout the Subscription Term.

Software updates may include:

• security improvements;
• bug fixes;
• AI model updates;
• workflow improvements;
• infrastructure optimizations;
• performance enhancements;
• new functionality.

Routine updates shall not require Customer approval.

13.2 Major Releases

Where a software release is reasonably expected to materially affect Customer's use of the Services, ONLYAI shall use commercially reasonable efforts to provide advance notice through the Customer Portal, electronic mail or other designated communication channels.

13.3 Security Releases

Security-related software updates may be deployed immediately where reasonably necessary to:

• remediate vulnerabilities;
• address active threats;
• comply with Applicable Laws;
• maintain operational integrity.

Customer acknowledges that such deployments may occur without prior notice where immediate action is reasonably required.

14. BUSINESS CONTINUITY AND DISASTER RECOVERY

14.1 Business Continuity

ONLYAI shall maintain commercially reasonable Business Continuity procedures designed to facilitate restoration of critical operations following a Business Continuity Event.

Business Continuity planning shall be periodically reviewed and updated as part of ONLYAI's operational governance program.

14.2 Disaster Recovery

ONLYAI shall maintain disaster recovery capabilities appropriate to the nature of the Services.

Disaster recovery measures may include:

• geographically redundant infrastructure where commercially appropriate;
• encrypted backups;
• infrastructure replication;
• configuration management;
• recovery procedures;
• operational documentation.

14.3 Recovery Objectives

Unless otherwise expressly specified in an applicable Order Form, ONLYAI shall maintain internal recovery objectives appropriate to the nature and criticality of the affected Services.

Any Recovery Time Objective or Recovery Point Objective communicated by ONLYAI shall constitute an operational target rather than a contractual guarantee and shall remain subject to:

(a) the nature and severity of the relevant Disaster Recovery Event;

(b) the availability and recovery capabilities of applicable Third-Party Dependencies;

(c) the integrity and availability of backups;

(d) technical dependencies outside ONLYAI’s reasonable control; and

(e) any other circumstances materially affecting recovery.

Specific contractual RTO or RPO commitments shall apply only where expressly agreed in the applicable Order Form.

14.4 Backup Management

ONLYAI shall maintain encrypted backups of production systems in accordance with its internal backup policy.

Backups are maintained solely for disaster recovery and operational resilience and are not intended to serve as Customer's sole archival solution.

Customer remains responsible for maintaining any separate archival or regulatory retention requirements applicable to its business.

15. PERFORMANCE MANAGEMENT

15.1 Performance Monitoring

ONLYAI shall maintain commercially reasonable monitoring of key operational performance indicators, which may include:

• infrastructure utilization;
• API performance;
• AI inference latency;
• workflow execution times;
• queue utilization;
• storage capacity;
• database performance.

Performance metrics are maintained for operational purposes and do not constitute contractual performance guarantees unless expressly set forth in an applicable Order Form.

15.2 Capacity Management

ONLYAI shall use commercially reasonable efforts to maintain sufficient production capacity to support subscribed Services under normal operating conditions.

Where Customer's usage materially exceeds anticipated operational parameters, ONLYAI may recommend migration to an alternative subscription tier or implementation of additional capacity planning measures.

16. SERVICE CREDITS

16.1 Exclusive Financial Remedy

Subject to the terms and conditions of this SLA, Service Credits shall constitute Customer's sole and exclusive financial remedy for any failure by ONLYAI to achieve the applicable Monthly Service Availability Commitment.

Service Credits shall not apply to any other obligations arising under the Agreement, including obligations relating to data protection, confidentiality, intellectual property, information security or indemnification.

For the avoidance of doubt, Service Credits shall not limit Customer's rights in respect of a material breach of the Agreement where such rights are expressly preserved under the Agreement.

16.2 Eligibility

Customer shall become eligible for Service Credits only where:

(a) the applicable Availability Commitment has not been achieved;

(b) the Downtime is not attributable to Excluded Downtime;

(c) Customer has complied with all obligations under the Agreement and this SLA;

(d) Customer submits a written Service Credit request within thirty (30) calendar days following the end of the relevant calendar month.

Failure to submit a timely request shall constitute a waiver of the applicable Service Credit.

For the avoidance of doubt, no Service Credit shall accrue in respect of any Excluded Downtime or any period during which the availability, performance or functionality of the Services is materially affected by a Third-Party Dependency or other circumstance outside ONLYAI’s reasonable control.

16.3 Service Credit Schedule

Subject to Section 16.2, Service Credits shall be calculated as follows:

For purposes of this Section, "Monthly Subscription Fee" means the recurring subscription fees payable under the applicable Order Form, excluding:

• Revenue Share;
• Professional Services;
• Taxes;
• Third-Party Charges;
• One-time implementation fees.

16.4 Application of Service Credits

Approved Service Credits shall:

• be applied against future invoices;
• not be redeemable for cash;
• not accrue interest;
• not be transferable;
• expire upon termination of the Agreement unless otherwise required by Applicable Law.

16.5 Maximum Service Credits

The aggregate Service Credits payable during any calendar month shall not exceed twenty percent (20%) of the Monthly Subscription Fee applicable to the affected Services.

17. EXCLUDED DOWNTIME

The Availability Commitment shall not apply to, and Downtime shall exclude, any unavailability, degradation, latency, interruption, restriction or loss of functionality resulting directly or indirectly from:

(a) Scheduled Maintenance;

(b) Emergency Maintenance;

(c) Force Majeure Events;

(d) any failure, outage, degradation, latency, restriction, rate limitation, modification, suspension or discontinuation affecting a Third-Party Dependency;

(e) any failure or degradation affecting a third-party artificial intelligence provider, model routing service, large language model provider, foundation model provider, inference provider or AI infrastructure provider;

(f) any failure, outage, degradation, modification, restriction, rate limitation, suspension or discontinuation affecting an external or third-party API;

(g) any failure, outage, degradation or interruption affecting a Supported Platform or any third-party service facilitating connectivity with a Supported Platform;

(h) any failure, outage, degradation or interruption affecting cloud infrastructure, hosting infrastructure, managed database services, object storage services, media storage services, content delivery networks, domain name services, reverse proxies, web application firewalls or distributed denial-of-service mitigation services;

(i) any failure, outage, degradation or interruption affecting queueing, caching, asynchronous task processing, messaging or event delivery infrastructure;

(j) any failure, outage, degradation or interruption affecting telecommunications providers, internet service providers, backbone networks, routing infrastructure or other communications networks;

(k) any failure or unavailability of third-party authentication, identity management, monitoring, logging, analytics, transactional communication or security services;

(l) Customer’s network infrastructure, internet connectivity, hardware or software;

(m) unsupported browsers or operating systems;

(n) Customer-developed integrations or third-party integrations selected, configured or introduced by Customer;

(o) Customer misuse of the Services;

(p) unauthorized modifications;

(q) excessive API requests or usage exceeding contractual, technical or documented limits;

(r) suspension of Services permitted under the Agreement;

(s) acts or omissions of Customer or its Authorized Users;

(t) security incidents caused or materially contributed to by Customer; or

(u) any other event or circumstance outside ONLYAI’s reasonable control.

For the avoidance of doubt, the fact that ONLYAI has selected, contracted with, integrated, configured or commercially relies upon a Third-Party Dependency shall not, by itself, cause an outage, failure or degradation of such Third-Party Dependency to constitute Downtime attributable to ONLYAI.

Where an event described in this Section affects the Services, ONLYAI shall use commercially reasonable efforts to mitigate the resulting impact and restore affected functionality to the extent technically and commercially practicable and within ONLYAI’s reasonable control.

18. CUSTOMER RESPONSIBILITIES

Customer acknowledges that achievement of the Service Levels established under this SLA depends upon Customer fulfilling certain operational responsibilities.

Accordingly, Customer shall:

• maintain appropriate internet connectivity;
• maintain supported browsers and operating systems;
• promptly install required updates where applicable;
• implement appropriate cybersecurity controls within Customer's own environment;
• maintain secure authentication credentials;
• designate qualified Authorized Support Contacts;
• cooperate in Incident investigations;
• promptly report suspected service disruptions;
• maintain accurate configuration information;
• ensure that Authorized Users receive appropriate training regarding the operation of the Services.

ONLYAI shall not be responsible for any degradation of service resulting from Customer's failure to satisfy the foregoing obligations.

19. PERFORMANCE REPORTING

19.1 Operational Reporting

ONLYAI may maintain internal operational metrics relating to:

• system availability;
• infrastructure health;
• API availability;
• support response performance;
• Incident statistics;
• AI service health;
• security monitoring;
• infrastructure utilization.

Such metrics shall be maintained primarily for operational management purposes.

19.2 Enterprise Reporting

Enterprise Customers may request periodic operational reports summarizing, where applicable:

• Service Availability;
• Incident history;
• maintenance activities;
• major software releases;
• security notifications;
• Service Credit calculations;
• operational trends.

The format and frequency of such reports shall be determined by ONLYAI in its reasonable discretion unless otherwise agreed in the applicable Order Form.

20. SERVICE GOVERNANCE

20.1 Governance Meetings

For Enterprise Customers, the Parties may conduct periodic Service Governance Meetings.

Such meetings may address:

• platform performance;
• operational metrics;
• roadmap updates;
• major incidents;
• recurring operational issues;
• security developments;
• regulatory developments affecting the Services;
• customer feedback.

Governance Meetings are intended to facilitate continuous operational improvement and shall not amend the Agreement unless expressly documented in writing.

20.2 Continuous Improvement

ONLYAI operates a continuous improvement program designed to enhance the Services over time.

Accordingly, ONLYAI may periodically implement:

• security improvements;
• AI model enhancements;
• infrastructure optimization;
• workflow improvements;
• performance enhancements;
• monitoring improvements;
• user interface improvements;
• operational automation.

Customer acknowledges that continuous improvement constitutes an essential characteristic of modern AI-enabled cloud services.

21. COMPLIANCE WITH INDUSTRY STANDARDS

ONLYAI shall use commercially reasonable efforts to maintain operational practices generally consistent with recognized industry standards applicable to enterprise cloud software providers.

Such practices may include, where appropriate:

• secure software development lifecycle (SSDLC);
• vulnerability management;
• change management;
• access management;
• logging and monitoring;
• incident response;
• business continuity planning;
• disaster recovery planning.

Nothing contained in this Section shall be interpreted as representing formal certification under any particular standard unless expressly stated by ONLYAI in writing.

22. MODIFICATIONS TO THIS SLA

ONLYAI may modify this SLA from time to time in order to:

• reflect technological developments;
• improve operational practices;
• address security requirements;
• respond to regulatory changes;
• improve service quality.

No modification shall materially diminish Customer's contractual rights during the then-current Subscription Term without Customer's prior written consent, except where such modification is required by Applicable Law or is necessary to address an imminent security risk.

23. RELATIONSHIP WITH THE AGREEMENT

This SLA forms an integral part of the Agreement.

Except as expressly modified herein, all provisions of the Agreement shall remain in full force and effect.

Nothing contained in this SLA shall:

• expand ONLYAI's liability;
• create additional warranties;
• modify the limitation of liability contained in the Agreement;
• create service obligations beyond those expressly set forth herein.

24. SURVIVAL

The following Sections shall survive termination or expiration of the Agreement to the extent necessary to give effect to their purpose:

• Service Credits accrued prior to termination;
• Confidentiality obligations relating to operational information;
• Dispute Resolution;
• Governing Law;
• Liability provisions incorporated from the Agreement.

25. ACCEPTANCE

This SLA is incorporated into the Terms of Service and becomes effective upon Customer’s electronic acceptance of the Agreement or electronic submission of the applicable Order Form, whichever occurs first, as recorded by ONLYAI’s systems. No signature is required.