# Pull That Up Jamie

> Pull That Up Jamie is a podcast search and research platform with a semantically indexed corpus of 109+ feeds, ~7K episodes, and ~1.9M paragraphs. It provides vector search across podcast transcripts, interactive research sessions, on-demand transcription, and clip creation — all accessible via a public API with free and Lightning-paid (L402) tiers.

## API

- [OpenAPI Specification](https://www.pullthatupjamie.ai/api/openapi.json): Full machine-readable API spec (OpenAPI 3.0) covering search, corpus discovery, research sessions, clip creation, on-demand transcription, and L402 agent auth
- [Corpus API Spec (Markdown)](https://www.pullthatupjamie.ai/api/corpus/spec): Human/LLM-readable markdown reference for the corpus navigation endpoints

## Agent Integration

- [ClawHub Skill](https://clawhub.ai/unclejim21/pullthatupjamie): Install as an agent skill — includes auth flow, search patterns, and research session workflows
- [Corpus Stats](https://www.pullthatupjamie.ai/api/corpus/stats): Live corpus-wide statistics (feeds, episodes, chapters, paragraphs, people, topics) — no auth required
- [Corpus Feeds](https://www.pullthatupjamie.ai/api/corpus/feeds): Browse all indexed podcast feeds — no auth required

## Key Endpoints

- [Semantic Search](https://www.pullthatupjamie.ai/api/search-quotes): Vector search across podcast transcripts with filtering by feed, episode, date range
- [Chapter Search](https://www.pullthatupjamie.ai/api/search-chapters): Search chapter metadata (keywords, headlines, summaries) across the corpus — L402 required
- [Podcast Discovery](https://www.pullthatupjamie.ai/api/discover-podcasts): LLM-assisted podcast discovery across 4M+ podcasts via the Podcast Index — free tier available
- [On-Demand Transcription](https://www.pullthatupjamie.ai/api/on-demand/submitOnDemandRun): Submit any podcast episode for transcription, chaptering, and permanent semantic indexing
- [Clip Creation](https://www.pullthatupjamie.ai/api/make-clip): Generate shareable audio/video clips with burned-in subtitles from search results
- [Transcript Download](https://www.pullthatupjamie.ai/api/on-demand/transcript/{jobId}/{guid}): Download full transcript JSON for a transcribed episode — requires the same L402 credential used to pay for the job (streamed, supports large files)

## Canonical Workflows

### Deep Topic Research
Survey the landscape, then drill into specifics, then collect and analyze.
1. `POST /api/search-chapters` with a broad topic query — returns chapter headlines, summaries, and timestamps across the corpus to see who's discussing the topic and how it evolved
2. `POST /api/search-quotes` with `smartMode: true` — deep dive into specific transcript segments using natural language (e.g. "What was Saylor saying about AI on TFTC last year")
3. `POST /api/research-sessions/` with collected pineconeIds — bundle findings into a research session
4. `POST /api/research-sessions/{id}/share` — create a shareable link with preview image
5. `POST /api/research-sessions/{id}/analyze` — stream an AI analysis of the collected clips

### Person Dossier / Meeting Prep
Find everything someone has said publicly across indexed podcasts.
1. `GET /api/corpus/people?search={name}` — find the person and their appearance count
2. `POST /api/corpus/people/episodes` with `{ "name": "..." }` — get all episodes they appeared on
3. `POST /api/search-quotes` filtered by their feedIds or episode GUIDs — search their quotes
4. `POST /api/research-sessions/` — collect relevant clips into a session
5. `POST /api/research-sessions/{id}/share` — share the dossier

### Discover → Index → Search (New Content)
Find a podcast or episode that isn't indexed yet, get it transcribed, then search it.
1. `POST /api/discover-podcasts` with a natural language query — returns matching podcasts/episodes across 4M+ feeds, each flagged with transcript availability
2. If `hasTranscripts: true` → go straight to `POST /api/search-quotes` with the feedId
3. If `hasTranscripts: false` → `POST /api/on-demand/submitOnDemandRun` with the episode GUID and feedId to submit for transcription
4. `POST /api/on-demand/getOnDemandJobStatus` with the jobId — poll every 15s until status is "complete" (typically 30-120s per episode). If you include the same L402 `Authorization` header used to pay, each completed episode includes a `transcriptUrl`
5. `GET /api/on-demand/transcript/{jobId}/{guid}` with the same L402 header — download the full transcript JSON
6. `POST /api/search-quotes` with the feedId — the newly transcribed content is now permanently searchable

### Quote → Clip → Share
Find an exact quote and turn it into a shareable video clip with burned-in subtitles.
1. `POST /api/search-quotes` — find the quote with timestamp and audio deeplink
2. `POST /api/make-clip` with the clipId from the search result — kicks off async clip generation
3. `GET /api/clip-status/{lookupHash}` — poll every 5s until status is "completed" (typically under 1 min)
4. Use the returned clip URL — an MP4 with burned-in subtitles ready to share

### Competitive Intelligence / Sentiment Tracking
Track how a topic or narrative evolves over time across the corpus.
1. `POST /api/search-chapters` with date filtering — survey chapter-level coverage across a time range
2. `POST /api/search-quotes` with `minDate`/`maxDate` filters — drill into transcript segments for specific periods
3. `POST /api/research-sessions/` — collect clips from different time periods
4. `POST /api/research-sessions/{id}/analyze` with custom instructions (e.g. "Compare how the narrative shifted between Q1 and Q3 2025")

### Open-Ended Research (Agent Starting from a Vague Goal)
When starting from a broad or ambiguous request like "research what people are saying about X".
1. `POST /api/discover-podcasts` with the vague query — find relevant podcasts and episodes across the full Podcast Index catalog
2. For each result, check the transcript availability flag:
   - **Transcript available** → `POST /api/search-quotes` with the feedId to search existing transcripts immediately
   - **No transcript** → `POST /api/on-demand/submitOnDemandRun` to request transcription, then poll `/api/on-demand/getOnDemandJobStatus` until complete (include L402 header to get `transcriptUrl` for full transcript download), then search
3. `POST /api/search-chapters` across discovered feedIds — broaden the survey to chapter-level metadata for additional context
4. `POST /api/research-sessions/` — collect the best findings into a session
5. `POST /api/research-sessions/{id}/analyze` — generate an AI synthesis of everything found
6. `POST /api/research-sessions/{id}/share` — deliver the interactive research session link as the final output

## Auth

- Free tier: corpus browsing, basic search (IP-based quota), and shared session viewing — no credentials needed
- Paid tier: L402 Lightning prepaid credits. Hit any paid endpoint unauthenticated to receive a 402 challenge with a Lightning invoice. After payment, use `Authorization: L402 <macaroon>:<preimage>` for all requests. 500 sats (~$0.33) covers ~150+ searches.

## About

- [Blog](https://pullthatupjamie.ai/blog): Articles about the platform and podcast search
- [For Podcasters](https://pullthatupjamie.ai/for-podcasters): Tools and features built for podcast creators

## Optional

- [Privacy Policy](https://pullthatupjamie.ai/privacy): Data protection and privacy information
- [Terms of Service](https://pullthatupjamie.ai/terms): Service terms