Developer TermsPlayer TermsAcceptable UseAI UsagePrivacyDPASecuritySubprocessors

FirstLook Artificial Intelligence Usage Policy

Last Updated: August 19, 2026

1. Purpose and Scope

This policy describes how FirstLook (firstlook.gg), a product of Pragma Platform, Inc. (“Pragma,” “we” or “our”), uses artificial intelligence, the third-party model providers involved, the data processed by AI features, and the controls available to customers. It applies to all AI functionality offered within the FirstLook platform and supplements, but does not replace, our Data Processing Agreement (“DPA”) and published subprocessor list. In the event of any conflict between this policy and a customer’s agreement with Pragma (including the DPA), the agreement governs. For clarity, this policy is provided for transparency and does not modify any agreement between Pragma and a customer.

2. AI Features

FirstLook uses artificial intelligence for optional customer facing product features, including:

  • Sentiment Analysis. Analyzes community and playtest feedback, such as messages from Discord channels a customer has chosen to sync, in order to group sentiment and behavior signals.

  • FirstLook Agent. An in-product assistant that responds to user-submitted prompts and retrieves information from the customer’s own FirstLook workspace when a user requests it.

  • Localization. Uses generative AI to translate content within FirstLook into a customer’s supported languages.

FirstLook does not develop or train its own foundation models. AI functionality is delivered through third-party model providers accessed via application programming interfaces (APIs), as described in Section 3.

3. AI Model Providers

The following providers deliver AI functionality within FirstLook. Each is disclosed on our published subprocessor list at firstlook.gg/legal/subprocessors.

  • OpenAI (United States). Generative-AI inference supporting Sentiment Analysis.

  • Anthropic (United States). Generative-AI tooling and inference supporting the FirstLook Agent.

  • Fireworks AI (United States). Generative-AI inference used for certain features within FirstLook.

Changes to the subprocessor list, including AI providers, are published with advance notification and are subject to any objection right provided in the customer’s DPA.

4. Training Restrictions

We do not use personally identifiable customer data to train or fine-tune generative artificial intelligence models, and we do not permit our model providers to do so. We access model providers via APIs, under contractual terms that prohibit those providers from using submitted data to train their models, and we have disabled optional data-sharing settings with those providers. Model providers may temporarily retain prompts and outputs for a short period solely for abuse monitoring and service-integrity purposes, subject to those contractual terms.

For the sake of clarity, the restriction set forth above does not apply to de-identified data that does not identify a customer or any individual, including without limitation, product usage data, telemetry, diagnostic data, system logs, and support interactions.

5. Data Processed by AI Features

  • Sentiment Analysis. Customers select which sources and channels are synced for analysis and therefore control what content is submitted. This feature is designed to transmit to the model provider only the content from those selected sources, together with the context required to analyze it, and not to attach player data fields from the customer’s FirstLook environment.

  • FirstLook Agent. The Agent processes user-submitted prompts and content. When a user requests it, the Agent may retrieve player data points held within that customer’s FirstLook environment. The Agent is designed to access only data already processed under the customer’s DPA and not to access data outside the customer’s environment.

  • Localization. This feature transmits the content submitted for translation, together with necessary context, to the model provider.

6. Customer Controls

AI features are optional and disclosed. Sentiment Analysis and the FirstLook Agent may each be disabled for a customer’s account, including disabled by default from the outset of an engagement, upon request.

7. Ownership of AI-Generated Output

As between the parties, output generated by a customer using AI features within a customer’s FirstLook environment belongs to the customer. As between Pragma and the customer, Pragma retains all rights in the AI features themselves, the underlying models, and the FirstLook platform.

8. Security and Compliance

AI features operate within the same security program that governs the FirstLook platform generally, as described in our security documentation and the FirstLook Data Processing Agreement. Further detail is available upon request via security@pragma.gg. Our design and operation of AI features is informed by applicable law and recognized industry practices and, except as expressly agreed in a customer’s contract, this policy does not constitute a commitment to any particular third-party framework, standard, or certification.

9. Changes to This Policy

We may update this policy as our AI features and provider relationships evolve, including as AI features are introduced, modified, or discontinued. Material changes affecting subprocessors are communicated through the advance-notification process described in Section 3.

10. Contact

Security and privacy inquiries: security@pragma.gg (Information Security Officer).

Contractual and DPA inquiries: your account representative.

Pragma Platform, Inc. · 1401 21st Street Suite R, Sacramento, CA 95811 · firstlook.gg