These terms govern how you can use the AI features in every Delalify product and in the Conjoin AI service.
These Delalify AI Terms ('AI Terms') supplement and form part of the Delalify Terms of Service (the 'General Terms'). These AI Terms apply whenever you use an AI feature in any Delalify product, including Beryl in the Delalify Apps Platform and Conjoin AI in the Conjoin Platform.
These AI Terms govern both Beryl and Conjoin AI, which share the same underlying AI infrastructure. Beryl provides AI capabilities inside Delalify's applications for end users, while Conjoin AI provides those capabilities through APIs to developers who build their own applications.
These AI Terms are effective as of October 13, 2026.
If any part of these AI Terms conflicts with the General Terms, these AI Terms prevail for AI-related features and services. In addition, if you use Conjoin AI through the Conjoin Platform, the Conjoin-Specific Terms apply to that use as well.
Delalify routes AI requests to third-party AI model providers, whose models process your prompts and generate the responses. When you use an AI feature, Delalify sends your data to one or more of these providers for that processing. Before your data reaches any third-party provider, Delalify applies its policy controls, its redaction rules, and its audit logging.
You are responsible for complying with the acceptable use policies of the AI model providers whose models you use through Delalify. Delalify keeps a current list of these providers, with links to the terms and policies that apply to each one, on the AI Provider Policies page. Delalify may change that list whenever it adds or removes a provider.
Delalify selects and integrates the providers that deliver its AI capabilities; however, Delalify is not an agent of, a partner of, or a joint venture participant with any third-party AI provider. Each provider operates independently under its own terms and policies.
AI model providers may update their terms, modify their models, or discontinue their services at any time. Delalify does not control those changes and is not responsible for any change that a provider makes to a model's behavior, to its availability, or to its terms. If a provider updates its acceptable use policies, the updated policies apply to your continued use of that provider through Delalify.
If a model that you configured becomes unavailable, Delalify may route your requests to an alternative model that meets the same capability requirements. Delalify will notify you of material changes to the available models through your account dashboard or the Conjoin Console.
Delalify does not use your data, including your prompts, your context data, and the outputs that AI features generate for you, to train, fine-tune, develop, or improve any AI model. This commitment applies to every AI feature in both Beryl and Conjoin AI.
Delalify's contracts with third-party AI providers restrict those providers from using customer data that they process through the Delalify service to train or improve their models. However, Delalify does not control how those providers handle data outside those contracts. We recommend that you review the data handling policies of each provider, which you can open from the AI Provider Policies page, so that you understand how that provider handles data.
Delalify's Data Processing Agreement governs how Delalify handles your data when you use AI services. Delalify processes the data that you submit to AI features in accordance with that agreement and with Delalify's Privacy Policy.
Conjoin AI Policy provides redaction patterns that you can configure and enable so that Conjoin AI Policy redacts sensitive data before your data reaches any third-party AI model provider. The redaction categories that you can choose from include names, email addresses, phone numbers, credit card numbers, social security numbers, and other sensitive data fields.
Redaction stays off until you configure it in your Beryl settings or in your Conjoin AI Policy configuration. If you leave redaction off, Delalify sends your data to AI model providers exactly as you provided it. Delalify is not responsible for any sensitive data that it sends to model providers while redaction is off or while your redaction configuration permits specific fields.
Delalify applies redaction patterns on a best-effort basis. Although Delalify maintains and updates redaction patterns for common categories of sensitive data, those patterns may not detect every instance of sensitive data in every format. You retain full responsibility for complying with the data protection regulations that apply to you, which includes deciding which data needs protection and verifying that your redaction configuration is appropriate for your regulatory requirements.
Delalify does not guarantee that the outputs of AI features, including drafts, summaries, scores, suggestions, analyses, images, and other content, are accurate, complete, or free from errors. AI outputs may contain factual inaccuracies, outdated information, or misleading content.
AI outputs may not be unique, because other customers may receive similar or identical outputs from the same model or from different models in response to similar prompts.
Delalify does not warrant that AI outputs are suitable for any specific purpose. You are solely responsible for evaluating whether AI outputs are appropriate for your intended use, which includes deciding which uses need human review.
Delalify advises against relying on AI outputs as the sole basis for any decision that affects health, safety, legal rights, financial obligations, or employment. Wherever an error in an AI output can cause harm, Delalify advises you to verify that output independently before you act on it.
Delalify does not indemnify you against any intellectual property claim that arises from AI outputs. Before you publish, distribute, or otherwise use an AI output, you are responsible for evaluating whether that output infringes the intellectual property rights of a third party.
You may set spending limits on AI usage in your account settings or in your Conjoin AI Policy configuration. When you reach a spending limit, Delalify blocks further AI requests until you raise the limit or until a new billing period begins.
Delalify enforces spending limits as a product feature; however, Delalify is not liable for any charges that you incur between the time that you exceed a configured limit and the time that Delalify enforces that limit. You are responsible for configuring and monitoring the budget controls that suit your usage patterns.
The restrictions below apply to the AI features in Beryl, in Conjoin AI, and in every other Delalify service.
If you store generated content in Conjoin Storage, the content restrictions in the Conjoin-Specific Terms apply to that content as well.
You may use AI features for the purposes below only with appropriate human oversight and professional supervision.
Beryl and Conjoin AI Workflows can take autonomous actions, such as sending messages or updating records, only when you explicitly configure them to do so. By default, an approval gate holds every action that modifies data or communicates with an external party until a person approves that action.
You are responsible for configuring the approval requirements that suit your use case and your regulatory environment. Delalify recommends that you keep human review in place for every action that affects patient care, financial transactions, legal communications, or employment decisions.
For every AI request that Beryl or Conjoin AI processes, Delalify logs the input, the output, the model that produced the output, and the policy decisions that Delalify applied, including any redactions. Delalify retains these logs according to your data retention settings. Until Delalify deletes them, you can review the logs for compliance purposes through your Delalify account or the Conjoin Console.
The audit logs include timestamps, user identifiers, and enough request metadata to support regulatory reporting.
When you use AI features, you are responsible for every duty in the list below.
Delalify may modify these AI Terms from time to time to reflect changes in AI services, in provider relationships, in legal requirements, or in regulatory developments. Before any material change that reduces your rights or increases your obligations takes effect, Delalify will give you at least thirty (30) calendar days' notice, either by email to the address on your account or through an announcement in your account dashboard.
If you continue to use AI features after the effective date of a change, you accept the modified terms. If you disagree with a change, you must stop using AI features before its effective date.
Our team can walk you through data handling, provider policies, and the obligations that you take on when you use AI features.