Transpara AI Addendum

AI Addendum

Addendum to the Transpara Master Software License Agreement — Version 2026.2 — Effective October 5, 2026

This AI Addendum (this "Addendum") supplements the Transpara Master Software License Agreement (the "MSA") and the Transpara Product Schedule, and applies whenever the Transpara product is licensed under an Order Form. Because "Transpara" is also the name of the contracting party, this Addendum says "the Transpara product" where the product alone is meant. Capitalized terms used but not defined here have the meanings set forth in the MSA or the Transpara Product Schedule.

This Addendum is necessary because Customer's use of the Transpara product includes artificial-intelligence functionality that operates through Customer's chosen AI Model. The purpose of this Addendum is to allocate, between Transpara and Customer, the rights and risks associated with that functionality, and to make clear what Transpara does and does not warrant. In the event of a conflict between this Addendum and the MSA or the Transpara Product Schedule, the order of precedence in MSA Section 2.3 controls (Order Form > AI Addendum > Product Schedules > DPA > AUP > MSA).

1. Defined Terms

"AI Output" means any output, response, completion, recommendation, classification, prediction, suggestion, alert, or other content produced by a Customer-Selected AI Model in response to a prompt, query, instruction, or other input submitted through the Transpara product's AI Functionality. AI Outputs include the output of both Seek Mode and Surface Mode.

"AI Functionality" means the AI orchestration capabilities of the Transpara product, including the prompt-construction layer, the integration with Customer-Selected AI Models, Seek Mode, Surface Mode, MCP Integration, and any related features made available by Transpara.

"Customer-Selected AI Model" has the meaning given in the Transpara Product Schedule. Each Customer-Selected AI Model is licensed by Customer directly from the applicable Customer-Selected AI Model Provider.

"Customer-Selected AI Model Provider" means the third-party provider of a Customer-Selected AI Model (for example, Anthropic, OpenAI, Google, Microsoft, Amazon Web Services, Meta, or any other third-party AI Model provider, or, in the case of a Customer-hosted open-source or proprietary AI Model, Customer itself in its capacity as AI Model operator).

"Hallucination" means an AI Output that is factually inaccurate, fabricated, or otherwise unreliable, including outputs that confidently assert false propositions, invent citations, or misrepresent the underlying source material. Hallucinations are an inherent risk of generative AI and are not within Transpara's control.

"High-Risk AI Use" means any use of the Transpara product that would be classified as "high-risk" under the EU Regulation 2024/1689 (the "EU AI Act"), the Colorado AI Act (C.R.S. Section 6-1-1701 et seq.), or any similar legislation; or any use of the AI Functionality to make or substantially influence decisions concerning eligibility for credit, insurance, housing, employment, education, healthcare, public benefits, criminal justice, or other consequential decisions affecting natural persons.

"Layer 1 Knowledge" or

"Transpara Knowledge" has the meaning given in the Transpara Product Schedule. Layer 1 Knowledge is Transpara Confidential Information and a trade secret of Transpara.

"Layer 2 Knowledge" or

"Customer Knowledge" has the meaning given in the Transpara Product Schedule. Layer 2 Knowledge is part of Customer Data.

"MCP Integration" means the Model Context Protocol (or successor protocol) interfaces by which a Customer-Selected AI Model exchanges context, prompts, and responses with the Transpara product. MCP Integration in the Transpara product is configured to allow read and analysis operations only and does not provide any capability to take action against control systems or operational technology.

"Seek Mode" means the user-initiated AI mode of the Transpara product, in which an Authorized User explicitly invokes the AI Functionality (for example, by asking a question or requesting an analysis) and receives an advisory response.

"Surface Mode" means the autonomous monitoring AI mode of the Transpara product, in which the AI Functionality runs in the background, monitors Customer Data, and surfaces alerts, observations, anomalies, or recommendations without explicit user invocation. Surface Mode outputs are advisory only; Surface Mode does not take any action against control systems or operational technology.

2. Bring-Your-Own-Model (BYOM)

2.1 Customer's Selection and Licensing.

Customer is solely responsible for: (a) selecting the Customer-Selected AI Model(s) used with the Transpara product; (b) entering into and complying with the agreement with each Customer-Selected AI Model Provider, including paying all fees owed to the provider; (c) obtaining and maintaining all API keys, authentication credentials, subscriptions, and access permissions necessary to use the Customer-Selected AI Model with the Transpara product; (d) complying with the Customer-Selected AI Model Provider's terms of service, acceptable use policy, data-handling commitments, and usage limits; and (e) ensuring that the Customer-Selected AI Model is appropriate for Customer's intended use, including with respect to accuracy, content filtering, data residency, AI Model training, output filtering, and applicable law.

2.2 Transpara as Orchestration Layer.

Transpara provides the orchestration layer that integrates the Transpara product with the Customer-Selected AI Model(s) configured by Customer. Transpara is not a Customer-Selected AI Model Provider, does not resell or distribute any Customer-Selected AI Model, is not a party to Customer's agreement with any Customer-Selected AI Model Provider, and does not warrant any Customer-Selected AI Model or any AI Output produced by it. The relationship between Customer and the Customer-Selected AI Model Provider is solely between them.

2.3 No Pass-Through of Provider Terms.

The terms of service, acceptable use policies, indemnification commitments, limitations of liability, content restrictions, and data-handling commitments of the Customer-Selected AI Model Providers flow between Customer and each provider, not through Transpara. Transpara makes no representations regarding those terms. Customer is responsible for vetting each Customer-Selected AI Model Provider's terms before configuring the AI Model for use with the Transpara product and for ensuring that those terms align with Customer's data-handling requirements.

2.4 Cost.

Customer pays all charges owed to the Customer-Selected AI Model Provider directly. Transpara does not invoice for, mark up, or pass through Customer-Selected AI Model usage charges. Transpara's Fees under the MSA cover the Transpara product and its orchestration layer only.

2.5 Configuration.

Customer configures which Customer-Selected AI Model is used for which portion of the Transpara product's AI Functionality. Customer may use different Customer-Selected AI Models for different functions (for example, one AI Model for Seek Mode and a different AI Model for Surface Mode, or different AI Models for different sites). Customer is responsible for the consequences of its configuration choices.

3. Two-Layer Knowledge Architecture

3.1 Layer 1 — Transpara Knowledge.

Layer 1 Knowledge consists of the proprietary prompts, system messages, instructions, best-practice content, orchestration logic, configurations, and other content authored by or for Transpara and embedded in the Transpara product. Layer 1 Knowledge represents significant Transpara investment in industrial-domain expertise. Layer 1 Knowledge is Transpara Confidential Information (whether or not marked) and is protected as a Transpara trade secret. Customer will not, and will not permit any Authorized User or third party to: (a) extract, copy, log, store, or otherwise capture Layer 1 Knowledge; (b) reverse engineer, decompile, or otherwise attempt to derive Layer 1 Knowledge from observation of the Transpara product, AI Outputs, or any other source; (c) instruct any Customer-Selected AI Model to reveal Layer 1 Knowledge (for example, by prompt injection or similar techniques); or (d) use any AI Output or any other portion of the Transpara product to train, fine-tune, or improve any competing product or service. Any unauthorized extraction of Layer 1 Knowledge is a material breach of this Agreement.

3.2 Layer 2 — Customer Knowledge.

Layer 2 Knowledge consists of content authored, supplied, or configured by Customer (for example, Customer's standard operating procedures, rules, guardrails, domain content, custom instructions, and reference materials) and made available to the Transpara product. As between the parties, Customer owns Layer 2 Knowledge. Customer grants Transpara a limited, non-exclusive, royalty-free license to access and use Layer 2 Knowledge solely as necessary to provide the Transpara product and Support Services. Customer is responsible for ensuring that Layer 2 Knowledge does not contain content Customer is not authorized to use, share, or process through a Customer-Selected AI Model.

3.3 Customer Inputs.

Customer Data submitted to the Transpara product's AI Functionality ("Customer Inputs") belongs to Customer. Transpara does not claim any ownership interest in Customer Inputs.

3.4 AI Outputs.

As between Transpara and Customer, Customer owns AI Outputs produced in response to its Customer Inputs and Layer 2 Knowledge. Customer's ownership of AI Outputs is subject to (a) any limitations imposed by the applicable Customer-Selected AI Model Provider's terms (which may, for example, deny intellectual property protection for AI Outputs, claim derivative-work rights, or impose conditions on the use of the output), (b) any pre-existing rights of third parties in materials referenced or reproduced in AI Outputs, (c) Transpara's ownership of the underlying Transpara product, the AI Functionality, and Layer 1 Knowledge, and (d) the restrictions on use of AI Outputs in Section 7. Customer is solely responsible for determining whether and how to use any AI Output.

4. No Action — Monitoring and Analysis Only

4.1 Boundary.

The Transpara product is a monitoring, visualization, and analysis platform. The AI Functionality of the Transpara product — including Seek Mode, Surface Mode, and MCP Integration — produces advisory output only. The Transpara product does not, and is not designed to, issue, transmit, or execute commands, instructions, set points, or other control signals to any control system, actuator, valve, breaker, programmable logic controller, distributed control system, SCADA system, manufacturing execution system, or other operational-technology system.

4.2 MCP Integration Boundaries.

Where MCP Integration is configured, the Transpara product exposes a constrained set of capabilities to the Customer-Selected AI Model. Those capabilities are limited to reading Customer Data, querying configurations and metadata, performing analysis, and producing advisory outputs. The Transpara product's MCP Integration does not expose any capability to write to control systems, issue operational commands, modify configurations of operational technology, or otherwise take action in the Customer Environment. Customer will not develop or deploy custom MCP servers, tools, or integrations that would extend the Transpara product's capability beyond monitoring and analysis.

4.3 Customer as the Actor.

Any action taken in the Customer Environment, including (without limitation) any change to set points, configurations, schedules, work orders, dispatch decisions, maintenance activities, or other operational matters, is taken by Customer (or its operators, technicians, or other personnel). The Transpara product's AI Functionality may inform such actions through advisory outputs, but the Transpara product does not take such actions and is not the proximate cause of any such action. Customer is solely responsible for: (a) reviewing AI Outputs for relevance, accuracy, and appropriateness before relying on them; (b) implementing appropriate human-in-the-loop processes for any action informed by the Transpara product; and (c) all consequences of any operational decision or action, whether or not informed by the Transpara product.

5. AI Output Accuracy; No Warranty

5.1 Probabilistic Nature.

Customer acknowledges that the AI Functionality is built on probabilistic generative-AI technology. AI Outputs are not deterministic. Different invocations of the same prompt may produce different responses. AI Outputs may contain Hallucinations, incorrect citations, outdated information, biased or unrepresentative content, or content that is factually inaccurate in ways that are not always detectable from the output alone. Customer is responsible for verifying AI Outputs before relying on them for any consequential decision.

5.2 Accuracy Depends on Customer-Selected AI Model.

The accuracy, completeness, reliability, and appropriateness of AI Outputs depend on (a) the Customer-Selected AI Model selected and configured by Customer, including its training, fine-tuning, and version, (b) the Customer Inputs and Layer 2 Knowledge supplied by Customer, (c) the prompt-construction logic in Layer 1 Knowledge, and (d) the underlying real-world phenomena being modeled. Transpara warrants its own orchestration layer and Layer 1 Knowledge only to the extent expressly set forth in this Addendum and the MSA. Transpara does not warrant the accuracy, completeness, reliability, or appropriateness of any AI Output for any particular purpose.

5.3 Disclaimer.

EXCEPT FOR THE EXPRESS WARRANTIES IN MSA Section 12.2 (WHICH APPLY TO THE TRANSPARA PRODUCT AS DELIVERED BUT DO NOT WARRANT ACCURACY OR OPERATIONAL OUTCOMES), TRANSPARA EXPRESSLY DISCLAIMS ALL WARRANTIES WITH RESPECT TO AI OUTPUTS, WHETHER EXPRESS, IMPLIED, OR STATUTORY, INCLUDING ANY WARRANTIES OF ACCURACY, COMPLETENESS, NON-INFRINGEMENT, FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR NON-HALLUCINATION.

6. Training and Data Retention

6.1 Transpara Does Not Train.

Transpara does not train, fine-tune, retrain, or otherwise use Customer Data (including Customer Inputs, Layer 2 Knowledge, and AI Outputs) to train any AI Model. Transpara is not the operator of any Customer-Selected AI Model. Transpara does not retain AI Outputs except as expressly authorized by Customer or as may incidentally occur during Support Services initiated by Customer.

6.2 Customer-Selected AI Model Provider May Train.

Customer acknowledges that, depending on Customer's agreement with each Customer-Selected AI Model Provider, the Customer-Selected AI Model Provider may use Customer Inputs and AI Outputs to train, fine-tune, or improve its AI Models, or may not. Whether or not the Customer-Selected AI Model Provider trains on Customer Data is governed solely by Customer's agreement with that provider. Customer is responsible for selecting Customer-Selected AI Models whose training behavior aligns with Customer's data-handling requirements (for example, by selecting an enterprise or no-training tier offered by the provider, by configuring the AI Model with no-training settings, or by hosting the AI Model in Customer's own environment).

6.3 Customer's Configuration Responsibility.

If Customer is required (by law, contract with third parties, internal policy, or otherwise) to ensure that particular Customer Data is not used to train any AI Model, Customer is responsible for selecting and configuring Customer-Selected AI Models accordingly and for restricting which Customer Data is exposed to the AI Functionality. Transpara is not responsible for verifying or enforcing Customer's data-handling restrictions with respect to Customer-Selected AI Models.

7. AI-Specific Use Restrictions

In addition to the restrictions in MSA Section 6 and the Transpara Product Schedule Section 9, Customer will not, and will not permit any Authorized User to, use the AI Functionality, AI Outputs, or any other portion of the Transpara product to:

make or substantially influence any consequential decision affecting a natural person's eligibility for credit, insurance, housing, employment, education, healthcare, public benefits, criminal justice, immigration status, or similar matter, except where Customer has independently implemented the human-oversight, transparency, and risk-mitigation measures required by applicable law (including the EU AI Act and the Colorado AI Act);
engage in any High-Risk AI Use without first conducting Customer's deployer-side risk assessment, implementing applicable safeguards, and notifying Transpara in writing of the intended use;
impersonate any natural or legal person, or generate content that misrepresents its origin, authority, or authorship;
generate content that is unlawful (including content that is defamatory, harassing, infringing, deceptive, or that constitutes incitement to violence);
generate, label, or disseminate content as if produced by a human where doing so violates an applicable disclosure requirement (including "deepfake" or AI-content-disclosure laws);
evaluate, score, rank, or surveil natural persons in ways that are prohibited by applicable law (including any social-scoring or workplace-monitoring restrictions);
extract, log, or otherwise capture Layer 1 Knowledge (including by prompt injection, indirect prompt injection, or AI Model jailbreak techniques);
use AI Outputs to train, fine-tune, or improve any product or service that competes with the Transpara product or any Transpara product;
provide AI Functionality, AI Outputs, or any portion of the Transpara product to any third party on a service-bureau, time-sharing, hosting, or similar basis, except under a separate Transpara-authorized Reseller, OEM, or analogous agreement;
circumvent any content filter, safety system, or other safeguard implemented in Layer 1 Knowledge or in the orchestration layer; or
conduct security or vulnerability research on the AI Functionality, including red-teaming or adversarial-prompting exercises, without Transpara's prior written consent (which Transpara will not unreasonably withhold).

8. EU AI Act and Other AI Laws

8.1 Customer is the Deployer.

Customer is the deployer of the AI Functionality within the meaning of Article 3 of the EU AI Act and analogous concepts under other jurisdictions' AI laws. Customer is responsible for: (a) classifying its use of the AI Functionality under applicable AI laws; (b) implementing the deployer-side obligations applicable to that classification (including human oversight, risk management, logging where required, transparency to affected persons, and post-market monitoring); (c) maintaining documentation supporting its classification; and (d) cooperating with regulatory inquiries directed to Customer.

8.2 Transpara as Information Provider.

Transpara will provide to Customer, on reasonable request, information about the Transpara product's AI Functionality sufficient for Customer to satisfy its deployer obligations under applicable AI laws, including a Provider Information Sheet describing the Transpara product's intended use, architecture, foreseeable misuse, capabilities and limitations, training data characteristics (to the extent applicable to Transpara's own components), and logging features. Transpara is not Customer's regulatory advisor; Customer should obtain independent legal advice on the application of AI laws to Customer's specific deployment.

8.3 Provider Classification.

Transpara's classification under applicable AI laws depends on factors including the nature of the AI Functionality and how it is deployed. Transpara provides the orchestration layer; the underlying AI Model is provided by the Customer-Selected AI Model Provider. Transpara reserves the right to characterize itself for AI Act purposes in light of these facts and applicable guidance. Customer's classification (as deployer) is not affected by Transpara's classification.

9. Indemnification (AI-Specific)

9.1 Carve-Out from Transpara IP Indemnity.

Notwithstanding MSA Section 13.1 (IP Indemnification by Transpara), Transpara has no obligation to defend or indemnify Customer against any third-party claim to the extent it arises from or relates to: (a) any AI Output produced by a Customer-Selected AI Model; (b) any Customer-Selected AI Model; (c) Customer's combination of the Transpara product with a Customer-Selected AI Model or any other product, service, or material not provided by Transpara; (d) Customer's selection of, or failure to vet, any Customer-Selected AI Model or Customer-Selected AI Model Provider; or (e) Customer's failure to comply with the terms of any Customer-Selected AI Model Provider. Transpara's IP indemnification under MSA Section 13.1 continues to apply to the Transpara product (including the orchestration layer and Layer 1 Knowledge) as delivered by Transpara and not subject to a carve-out in this Section.

9.2 Customer Indemnification.

Customer will defend, indemnify, and hold harmless Transpara and its Affiliates from and against any third-party claim arising from or relating to: (a) any AI Output (including any claim that an AI Output is defamatory, infringes intellectual property, violates privacy, or is otherwise unlawful); (b) Customer's use of any Customer-Selected AI Model, including any breach of the Customer-Selected AI Model Provider's terms; (c) any Customer-Selected AI Model's training, output behavior, or intellectual property issues at the AI Model level; (d) Customer's classification or implementation of the AI Functionality under applicable AI laws; (e) any High-Risk AI Use; and (f) any breach by Customer of Section 7 of this Addendum (AI-Specific Use Restrictions). The indemnification procedure in MSA Section 13.5 applies.

10. Logging and Records

The Transpara product may maintain logs of prompts, AI Outputs, configurations, errors, and other operational events in the Customer Environment. Such logs are part of Customer Data and remain in the Customer Environment under Customer's control. Customer is responsible for the retention, security, and disposition of such logs in accordance with Customer's policies and applicable law. Transpara does not have access to Customer's AI Functionality logs except as may be necessary to perform Support Services initiated by Customer.

11. Beta and Experimental AI Features

From time to time, Transpara may make available beta, preview, experimental, or early-access AI features ("Beta AI Features"). Beta AI Features are provided on an as-is basis, are excluded from the warranty in MSA Section 12.2 and the IP indemnification in MSA Section 13.1, may produce a higher rate of Hallucinations or other errors, and may be modified or discontinued at Transpara's discretion. Customer's use of Beta AI Features is at Customer's sole risk and is subject to any additional terms Transpara provides at the time of enrollment.

12. Survival

Sections 1, 3, 4, 5, 6, 7, 9, 10, and 11 of this Addendum survive termination or expiration of the Agreement. The disclaimers, indemnifications, and limitations of liability in this Addendum survive as needed to give them full effect.