3.1.1 AI & Automation — PraismaGPT — Overview

PraismaGPT is the platform’s integrated AI layer designed to assist users across content creation, analysis, communication, and automation workflows. It operates as an augmentative system: enhancing speed and consistency while keeping users in control of decisions, outputs, and approvals.

Purpose and Positioning

PraismaGPT is embedded directly into existing modules rather than exposed as a standalone tool. This ensures that AI assistance appears contextually where work is already happening.

Primary objectives:

Reduce repetitive manual effort

Improve consistency and quality

Accelerate drafting and analysis tasks

Support multilingual and cross-domain workflows

PraismaGPT does not act autonomously; all outputs require user initiation and confirmation.

Capability Domains

PraismaGPT provides assistance across multiple functional domains.

Supported domains include:

Content drafting and rewriting

Summarization and extraction

Translation and localization support

Insight generation from structured data

Context-aware suggestions in workflows

Capabilities are enabled selectively per module and feature.

Context Awareness

AI responses are generated using scoped contextual input derived from the active module, user permissions, and current task state. Context is explicitly bounded to prevent unintended data exposure.

Context sources may include:

Active document or message

Metadata and structured fields

User-selected inputs

Global system data is never implicitly injected.

Invocation Model

PraismaGPT is invoked explicitly through user actions such as prompts, buttons, or assisted fields. There is no background execution without user intent.

Example invocation:

$aiResponse = PraismaGPT::prompt([ 'context' => $document, 'instruction' => 'summarize' ]);

Output Handling

AI-generated output is returned as draft material. It is never auto-published, auto-sent, or auto-applied.

Output characteristics:

Clearly labeled as AI-generated

Editable before use

Traceable to the initiating action

Users remain fully responsible for final content.

Governance and Limits

PraismaGPT operates within defined usage limits enforced at tenant and plan level. Usage is tracked for transparency and cost control.

Governance principles:

Explicit quotas and rate limits

No training on tenant data

No cross-tenant context sharing

Usage metrics are visible to administrators.

Reliability and Fallback

If AI services are unavailable or rate-limited, core platform functionality remains unaffected. AI assistance degrades gracefully without blocking user workflows.

Security and Privacy

All AI interactions are tenant-scoped. Sensitive data is excluded by default and included only when explicitly selected by the user. Requests and responses are logged for audit without storing model internals.

Access to PraismaGPT features is controlled through role-based permissions.