diff --git a/docs-site/docs/features/podcast-maker/api-reference.md b/docs-site/docs/features/podcast-maker/api-reference.md
new file mode 100644
index 00000000..e530de19
--- /dev/null
+++ b/docs-site/docs/features/podcast-maker/api-reference.md
@@ -0,0 +1,93 @@
+# Podcast Maker API Reference
+
+Base prefix: `/api/podcast`
+
+This page summarizes the Podcast Maker endpoints currently represented in frontend and backend code.
+
+## Endpoints by workflow stage
+
+### Analysis and idea shaping
+
+- `POST /idea/enhance`
+- `POST /analyze`
+- `POST /regenerate-queries`
+
+### Research
+
+- `POST /research/exa`
+
+### Scripting
+
+- `POST /script`
+- `POST /script/approve`
+
+### Audio
+
+- `POST /audio/upload`
+- `POST /audio`
+- `POST /combine-audio`
+- `GET /audio/{filename}`
+
+### Images
+
+- `POST /image`
+- `GET /images/{path}`
+
+### Video
+
+- `POST /render/video`
+- `POST /render/combine-videos`
+- `GET /videos`
+- `GET /videos/{filename}`
+- `GET /final-videos/{filename}`
+
+### Avatars
+
+- `POST /avatar/upload`
+- `POST /avatar/make-presentable`
+- `POST /avatar/generate`
+
+### Projects
+
+- `POST /projects`
+- `GET /projects`
+- `GET /projects/{project_id}`
+- `PUT /projects/{project_id}`
+- `DELETE /projects/{project_id}`
+- `POST /projects/{project_id}/favorite`
+
+### Dubbing (backend available)
+
+- `POST /dub/audio`
+- `GET /dub/{task_id}/result`
+- `GET /dub/audio/{filename}`
+- `POST /dub/estimate`
+- `GET /dub/languages`
+- `GET /dub/voices`
+- `POST /dub/voices/clone`
+- `GET /dub/voices/{task_id}/result`
+- `GET /dub/voices/audio/{filename}`
+
+## Implementation details
+
+### Endpoint usage in frontend service
+
+The current `podcastApi.ts` directly calls these podcast routes for analysis, research, script, audio, image, video, avatar, and project workflows.
+
+Known gap:
+
+- `cancelTask()` is a placeholder that posts to `/api/story/task/{taskId}/cancel` rather than a dedicated podcast route.
+
+### Request/response model notes
+
+At a high level:
+
+- Script endpoints exchange `idea`, `duration_minutes`, `speakers`, and optional `research`/`analysis`/`bible` context.
+- Audio endpoints exchange scene identifiers, text, and voice/rendering options.
+- Video endpoints exchange scene identifiers plus `audio_url` and optional image/prompt context.
+- Project endpoints exchange project-level state payloads suitable for restoring workflow progress.
+
+## Engineering references
+
+- `docs/Podcast_maker/AI_PODCAST_BACKEND_REFERENCE.md`
+- `docs/Podcast_maker/PODCAST_PERSISTENCE_IMPLEMENTATION.md`
diff --git a/docs-site/docs/features/podcast-maker/implementation-overview.md b/docs-site/docs/features/podcast-maker/implementation-overview.md
new file mode 100644
index 00000000..59fa4069
--- /dev/null
+++ b/docs-site/docs/features/podcast-maker/implementation-overview.md
@@ -0,0 +1,60 @@
+# Podcast Maker Implementation Overview
+
+This page keeps implementation details in one place for engineering and advanced troubleshooting.
+
+## Architecture
+
+Podcast Maker is split into:
+
+- **Frontend orchestration service**: `frontend/src/services/podcastApi.ts`
+  - Coordinates step flow (analysis → research → script → audio/video)
+  - Runs preflight checks before expensive calls
+  - Maps API payloads into UI-friendly objects
+- **Backend podcast handlers**: `backend/api/podcast/handlers/*.py`
+  - Route-level APIs for analysis, research, script, media, and projects
+  - Authenticated operations with user-scoped media/project data
+
+## Frontend orchestration responsibilities
+
+Primary responsibilities in `podcastApi.ts`:
+
+- Create project analysis payloads and map response into Podcast Analysis UI data.
+- Build/validate research query payloads for Exa research route.
+- Generate script scenes and normalize scene/line structure for editor state.
+- Render per-scene audio and combine scenes into final audio.
+- Trigger scene image and video generation workflows.
+- Persist project state via project CRUD endpoints.
+
+## Backend handler modules
+
+- `analysis.py`: idea enhancement, analysis, regenerate-queries.
+- `research.py`: Exa research endpoint.
+- `script.py`: script generation and scene approval.
+- `audio.py`: audio upload, generation, combine, serving audio files.
+- `images.py`: scene image generation and image serving.
+- `video.py`: scene video generation, video listing/serving, combine videos.
+- `avatar.py`: avatar upload, avatar generation, avatar cleanup/presentability.
+- `projects.py`: create, get, update, list, delete, favorite project records.
+- `dubbing.py`: dubbing/voice clone lifecycle endpoints (currently backend-available).
+
+## Data models (functional view)
+
+At feature level, the flow revolves around:
+
+- **Project metadata**: `project_id`, idea, duration, speakers, budget and status fields.
+- **Analysis output**: audience, content type, keywords, outlines, title suggestions.
+- **Research output**: source list, summarized insights, fact cards for script grounding.
+- **Script output**: scenes with IDs, durations, emotions, and speaker lines.
+- **Media output**: audio files, scene images, scene videos, combined episode artifacts.
+
+## Operational notes
+
+- Preflight checks are used to fail fast on plan/credit constraints.
+- Some operations are synchronous (analysis/script/audio/image), while video is async task-based.
+- Client-side task polling is used for long-running jobs.
+
+## Engineering references
+
+- `docs/Podcast_maker/AI_PODCAST_BACKEND_REFERENCE.md`
+- `docs/Podcast_maker/PODCAST_API_CALL_ANALYSIS.md`
+- `docs/Podcast_maker/PODCAST_PLAN_COMPLETION_STATUS.md`
diff --git a/docs-site/docs/features/podcast-maker/overview.md b/docs-site/docs/features/podcast-maker/overview.md
new file mode 100644
index 00000000..90aec19a
--- /dev/null
+++ b/docs-site/docs/features/podcast-maker/overview.md
@@ -0,0 +1,57 @@
+# Podcast Maker Overview
+
+Podcast Maker helps you turn a topic idea into a polished episode draft with research, script generation, AI voice narration, and optional video scenes.
+
+## What you do in the product
+
+1. **Start with an idea** and episode settings (duration, speakers, style).
+2. **Review AI analysis** suggestions (audience fit, outline ideas, titles, takeaways).
+3. **Run research** from selected queries and use source-backed fact cards.
+4. **Generate and edit a script** scene-by-scene.
+5. **Generate voice audio** for each scene and combine clips into one episode file.
+6. **Optionally create scene images and talking-head videos**.
+7. **Save and revisit projects** from your episode/project list.
+
+## What you see in the UI
+
+- Suggested outlines, titles, and hooks after analysis.
+- A query approval step before research runs.
+- Fact cards and summarized research insights.
+- Scene-based script editor with approval actions.
+- Audio generation controls (voice, emotion, speed, format-related options).
+- Video task progress and completed video listing.
+- Project persistence (save/load/list/favorite/delete).
+
+## Feature status matrix (based on current code)
+
+| Capability | Status | Notes |
+|---|---|---|
+| Idea enhancement + analysis suggestions | **Implemented** | Frontend calls `/api/podcast/idea/enhance` and `/api/podcast/analyze`; backend handlers exist. |
+| Research with Exa flow | **Implemented** | Frontend uses `/api/podcast/research/exa`; backend Exa research route is present. |
+| Script generation + scene approval | **Implemented** | Frontend uses `/api/podcast/script` and `/api/podcast/script/approve`; backend handlers exist. |
+| Scene audio generation + combine audio | **Implemented** | Frontend uses `/api/podcast/audio` and `/api/podcast/combine-audio`; backend handlers exist. |
+| Scene image generation | **Implemented** | Frontend uses `/api/podcast/image`; backend image handler exists. |
+| Scene video generation + status polling + combine videos | **Implemented** | Frontend uses `/api/podcast/render/video`, `/api/podcast/task/{id}/status`, `/api/podcast/render/combine-videos`; backend video routes are present. |
+| Project CRUD + favorites | **Implemented** | Frontend calls `/api/podcast/projects*`; backend create/get/update/list/delete/favorite routes exist. |
+| Avatar upload/generate/make-presentable | **Implemented** | Frontend calls `/api/podcast/avatar/*`; backend routes exist. |
+| Audio dubbing + voice clone routes | **Partial** | Backend dubbing routes exist; not wired in `podcastApi.ts` yet. |
+| Task cancellation from Podcast Maker UI | **Partial** | Frontend has `cancelTask()` placeholder using `/api/story/task/.../cancel`, not a dedicated podcast cancel API path. |
+| Multi-provider research toggle in podcast service | **Planned/Not active in current frontend** | Podcast frontend currently targets Exa route directly instead of a user-facing provider switch in this API layer. |
+
+## Advanced / developer notes
+
+Most users can ignore this section.
+
+- Podcast Maker uses preflight checks before expensive operations (analysis/script/audio/research) to surface plan/credit issues early.
+- The frontend normalizes snake_case API responses into camelCase for UI components where needed.
+- Long-running video operations are task-based and polled from the client.
+
+## Engineering references
+
+These are internal planning/reference docs retained as source material:
+
+- `docs/Podcast_maker/AI_PODCAST_BACKEND_REFERENCE.md`
+- `docs/Podcast_maker/AI_PODCAST_ENHANCEMENTS.md`
+- `docs/Podcast_maker/PODCAST_API_CALL_ANALYSIS.md`
+- `docs/Podcast_maker/PODCAST_PERSISTENCE_IMPLEMENTATION.md`
+- `docs/Podcast_maker/PODCAST_PLAN_COMPLETION_STATUS.md`