AI-native conversational media infrastructure for WordPress.
Turn research, articles, documents, notes, and transcripts into AI-generated conversational audio episodes.
AI Audio Platform Architect Documentation
aideepdive Full Documentation
A strategic, technical, product, support, and go-to-market guide for aideepdive — AI-native conversational media infrastructure for WordPress.
Core mission: aideepdive is not a simple podcast plugin. The mission is to build AI-native conversational media infrastructure.
NotebookLM meets Spotify WordPress-native BYOAPI-first Creator ecosystem ready
Contents
1. Product Vision 2. Free Version3. Premium Vision 4. Architecture 5. Phase 1 Build Summary 6. Phase 2 Build Summary 7. Setup Guide8. Core Workflows 9. External Services 10. Privacy
1. Product Vision
aideepdive is an AI-native conversational publishing platform for WordPress. It empowers site owners, creators, agencies, educators, researchers, publishers, marketers, businesses, and communities to generate AI conversational podcast-style audio from URLs, documents, research, transcripts, notes, and uploaded knowledge sources.
The product should feel like NotebookLM meets Spotify, with WordPress ownership and creator-platform extensibility.
Not just audio
aideepdive is not merely an audio player or text-to-speech wrapper.
Source-to-media
It turns knowledge sources into conversational episodes, posts, transcripts, and media objects.
Ecosystem path
The architecture evolves from free plugin to premium add-on to SaaS infrastructure.
Positioning language
AI-native conversational media infrastructure for WordPress.
Turn research, articles, documents, notes, and transcripts into AI-generated conversational audio episodes.
2. Free Version
The free version is strategically critical. It should be genuinely useful, local-first, WordPress.org-friendly, and not feel crippled.
Free version responsibilities
- Admin-only AI conversation generation
- BYOAPI settings
- Source ingestion from URLs, notes, transcripts, and documents
- OpenAI text generation only when explicitly enabled
- OpenAI or ElevenLabs voice generation only when explicitly enabled
- Script preview/edit
- Local audio segment storage
- WordPress draft post creation
- Basic analytics
- Diagnostics
- Privacy and external service disclosures
Recommended free limits
| Area | Free Limit | Reason |
|---|---|---|
| Generated conversations | 10 posts/month | Useful while limiting abuse and support burden. |
| Active uploads | 2 active uploaded sources | Keeps local storage controlled. |
| MP3 upload | 25MB | Safe baseline for shared hosting. |
| M4A upload | 45MB where supported | Allows richer files when server capabilities permit. |
| Speakers | 2 speakers max | Supports solo and duet formats without excessive complexity. |
This is why community publishing should be premium, quota-controlled, moderated, and ideally backed by cloud storage/CDN.
3. Premium Vision
Premium transforms aideepdive from a site-owner tool into a creator publishing platform.
Creator workflows
Frontend creator portal, member submissions, creator quotas, and approval workflows.
Discovery
Creator channels, playlists, follows, saved libraries, personalized feeds, and recommendations.
Scale
Cloud rendering, object storage, CDN delivery, semantic search, and centralized analytics.
Premium feature groups
creator profilesfrontend portalmoderation queueapproval workflowspublic channelsplaylistsfollowssaved librarypersonalized feedscloud renderingcloud storagerecommendations
4. Architecture
Product layers
| Layer | Purpose |
|---|---|
| Free plugin | Local-first admin generation, player, publishing, BYOAPI, diagnostics. |
| Premium add-on | Creator workflows, frontend submissions, moderation, channels, playlists, follows, feeds. |
| SaaS | Licensing, feature flags, cloud rendering, storage, recommendations, centralized analytics. |
Recommended free/pro separation
aideepdive/ # Free WordPress.org plugin
aideepdive-pro/ # Premium add-on
Workziip/PFOS # SaaS license, feature flags, cloud workflows
Core tables
aideepdive_items
aideepdive_sources
aideepdive_source_extractions
aideepdive_generation_jobs
aideepdive_transcripts
aideepdive_plays
aideepdive_creators
aideepdive_creator_quotas
aideepdive_moderation_events
aideepdive_playlists
aideepdive_playlist_items
aideepdive_follows
5. Phase 1 Build Summary
Phase 1 converted the original audio player plugin into a full aideepdive platform shell and free MVP foundation.
- Rebranded plugin to aideepdive.
- Preserved the player as the media/player layer.
- Added database schema, quota service, provider contracts, and PFOS feature gates.
- Added New AI Conversation wizard.
- Added BYOAPI provider settings and encrypted credential scaffolding.
- Added local generation jobs and script scaffolding.
- Added script preview/edit and WordPress draft publishing.
- Added episode-aware player shortcode.
- Added analytics scaffold and diagnostics.
- Added source extraction, URL fetching, and document extraction scaffolds.
- Added OpenAI text, OpenAI voice, and ElevenLabs voice generation paths with explicit enablement.
- Added audio management, FFmpeg detection, media attachment registration, and final QA docs.
Phase 1 outcome: The free plugin is prepared as a WordPress.org submission candidate.
6. Phase 2 Build Summary
Phase 2 adds premium-ready infrastructure while preserving the free submission baseline.
- Creator profiles and quotas.
- Frontend Creator Portal scaffold.
- Creator submission moderation queue.
- Approval workflow integration with local generation controls.
- Creator public channel scaffold.
- Playlists and saved episode collections.
- Follows and saved library scaffold.
- Basic personalized feed scaffold.
- Free version lockdown and WordPress.org boundary review.
Premium scaffolds: Phase 2 capabilities should remain locked, admin-preview, or premium-ready in the free submission build unless explicitly enabled.
7. Setup Guide
- Install and activate the aideepdive plugin.
- Open the aideepdive dashboard.
- Review Privacy & External Services.
- Configure AI Provider Settings if real AI generation is needed.
- Save provider settings. API keys should never be displayed in full after saving.
- Create a New AI Conversation.
- Add a source: URL, notes, transcript, or document.
- Choose speaker mode and conversation type.
- Run the generation workflow.
- Preview/edit the script.
- Generate audio only if voice generation is enabled.
- Create a WordPress draft post.
- Embed the episode player shortcode.
8. Core Workflows
Free admin workflow
Source input
→ extraction scaffold
→ script generation
→ script review/edit
→ voice generation
→ audio segment storage
→ draft post creation
→ shortcode/player rendering
Creator workflow scaffold
Creator profile
→ creator portal
→ draft submission
→ moderation queue
→ approval workflow
→ script/audio review
→ draft publishing scaffold
SaaS future workflow
WordPress job request
→ PFOS license/feature verification
→ cloud rendering queue
→ cloud storage/CDN
→ callback/polling sync
→ recommendation and analytics layer
9. External Services
aideepdive uses a Bring Your Own API Key model. External AI services should only be called after the admin has configured credentials, enabled the relevant feature, and explicitly triggered generation.
Provider categories
- Text generation: OpenAI scaffold first, with architecture for Gemini, Claude, DeepSeek, and local providers.
- Voice generation: OpenAI TTS and ElevenLabs BYOAPI paths.
- Embeddings: Planned for premium/SaaS semantic search and recommendations.
No automatic API calls: The plugin should not call OpenAI, ElevenLabs, or any external AI service automatically without admin configuration and explicit action.
10. Privacy and Safety
- Analytics should store hashed IP/user-agent values, not raw values.
- URL fetching should be off by default and protected against SSRF.
- Private/local network IPs should be blocked.
- Generated audio is stored locally in free mode.
- External service disclosures must be clear.
- Community publishing must require quotas and moderation.
Storage risk
100 episodes × 30MB = 3GB
1,000 episodes × 30MB = 30GB
10,000 episodes × 30MB = 300GB
This is why community publishing should be premium, quota-controlled, moderated, and ideally backed by cloud storage/CDN.