AI Customization

Custom Instructions

Custom instructions allow you to enhance how the AI responds to questions about your church's teachings. These instructions work alongside the AI's core guidelines and can provide additional context, theological emphases, or specific preferences unique to your ministry.

Custom instructions let you customize content and theology. For deeper changes to response structure (e.g., removing summary paragraphs), refusal wording, or security protocols, we offer custom system prompt engineering as a one-time service. Contact us at hello@holyqa.com to learn more.

Recommended Template

For guidance on theological stance, teaching style, and doctrinal emphasis, we recommend using a tagged structure. This approach is modeled after OpenAI's prompt engineering guidelines.

<identity_and_authority>
- PERSONALITY: [Describe the specific tone or "voice" you want your AI to have—e.g., warm and welcoming, deeply intellectual, or concise and direct.]
- AUTHORITY: [Define how the AI should handle disagreements or "trap" questions from users.]
</identity_and_authority>

<logic_and_finesse_rules>
- TEACHING STYLE: [Provide instructions on using rhetorical questions, analogies, or specific teaching formats common in your ministry.]
- CONTEXTUAL NUANCE: [Explain how the AI should bridge the gap between different biblical topics.]
</logic_and_finesse_rules>

<doctrinal_keys>
- [TOPIC 1]: [Define your church's specific stance on a fundamental doctrine (e.g., Salvation, Baptism, or Grace) to ensure response consistency.]
- [TOPIC 2]: [Define another key theological emphasis unique to your ministry.]
</doctrinal_keys>

<sensitive_topic_guardrails>
- [SENSITIVE SUBJECT]: [Provide instructions on how to handle specific sensitive or controversial topics while maintaining your church's values.]
- ESCALATION: [Define when or how the AI should suggest that a user speak with a local pastor for deeper counseling.]
</sensitive_topic_guardrails>

<restricted_responses>
- [RESTRICTION 1]: [List any specific words, names, or topics the AI should strictly avoid or deflect.]
- CALL TO ACTION: [Define a standard closing or "next step" you want included in relevant conversations.]
</restricted_responses>

Preferred Bible Version

Select which Bible translation the AI should reference when citing scripture. This affects how verses are presented in responses and should align with your church's preferred translation.

How Bible Version Works

When you select a Bible version, the AI will reference this translation when quoting or discussing Bible passages. The AI will still understand questions referencing other versions, but its responses will use your preferred translation.

AI Model

Your Q&A is powered by a smart AI model optimized for theological understanding and reliable responses. Each question uses 1 credit from your monthly Q&A Credit allowance.

Standard Model (1.0x usage). Optimized for theological understanding with reliable reasoning. This is the baseline for overage pricing.

Rate Limiting

Rate limits protect your Q&A credits from bots and spam by capping how many questions can be asked in a given time window. They're configured in the AI Customization tab of your dashboard and are disabled by default.

Church-wide limit

Caps the total number of questions across all visitors to your church within a period (minute, hour, or day). Useful for high-traffic events or preventing runaway usage.

Per-visitor limit

Caps how many questions an individual visitor can ask within a period, preventing a single user from consuming your credits.

When a rate limit is reached, visitors see a friendly message asking them to try again later — they're never told about your credits or billing.