Last Updated: April 2026
38Hub integrates artificial intelligence throughout the platform to help creators produce better content more efficiently. We believe in being transparent about how AI is used, what data is involved, and what limitations exist. This page explains our AI practices in plain language.
We are committed to responsible AI use and will update this page as our AI capabilities evolve.
AI powers several core features of the 38Hub platform. Every AI feature is designed to assist you — not replace your creative judgment.
AI evaluates content ideas on a 1-10 scale based on relevance, virality potential, and audience fit, with a written rationale for each score.
AI generates drafts for social media posts, threads, articles, scripts, and other formats based on your ideas, notes, and instructions.
AI analyzes text, PDFs, and other sources to extract actionable content ideas, saving you hours of manual research.
AI creates images, thumbnails, and visual content based on text prompts and content context for use in your posts and articles.
AI generates diagrams, flowcharts, and other structured visual content to enhance your articles and educational content.
AI provides real-time suggestions for tags, categories, titles, and content improvements as you work within the platform.
38Hub integrates with multiple AI providers. All AI infrastructure is managed by 38Hub — you do not need to manage API keys or external accounts.
Anthropic (Claude)
Content writing, idea scoring, idea extraction, text analysis, and general-purpose AI tasks. Claude is the default provider for most text-based AI features.
OpenAI (GPT-4o)
Alternative AI provider for content writing, scoring, and extraction. Available as a user-selectable option for those who prefer OpenAI models.
Google (Gemini)
Visual generation (thumbnails, social posts, carousels, diagrams, logos), YouTube content analysis, and specialized text tasks. Also provides OAuth authentication and YouTube API integration.
OpenRouter
Aggregator that exposes additional models (Llama, Mistral, Qwen, and more). Selectable per-task by admins or per-user via Settings → Security.
Per-task model configuration:Workspace admins can pin every AI workflow (scoring, extraction, cooking, keyword research, etc.) to a specific provider and model in Admin → AI Models. This lets compliance-sensitive organisations route every call to a single approved vendor.
Bring Your Own Keys (BYOK):Lifetime-deal customers and enterprise teams can route AI calls through their own provider accounts. When BYOK is enabled, the platform never falls back to its own keys — calls fail with a 402 error rather than billing the platform pool.
Failure logging:When an AI call returns a terminal error, we log the task, provider, model, error code, and request ID to an admin-only table so issues can be triaged. The original prompt content is not stored in this log.
Important Notice
AI-generated content is provided as a creative aid and starting point. It should not be treated as legal, medical, financial, or professional advice. All AI-generated content may contain factual errors, inaccuracies, outdated information, or biases. You are responsible for reviewing, verifying, and editing all AI outputs before publishing or relying on them.
You are always in control of how AI is used with your content:
38Hub does not develop or train its own AI models. All AI functionality is delivered via API integrations with third-party providers (Anthropic, OpenAI, Google). Your content is not used to train, fine-tune, or improve 38Hub's own models because we do not have any.
When your content is sent to AI providers via their APIs, it is subject to each provider's data handling policies. As of our last review:
We recommend reviewing each provider's current terms for the most up-to-date information.
We are transparent about the inherent limitations of AI technology: