PullThatUpJamie ClawHub Skill: Verifiable Podcast Intel + The Machine Payable Web in ActionPullThatUpJamie ClawHub Skill: Verifiable Podcast Intel + The Machine Payable Web in Action
PullThatUpJamie is now live on ClawHub:
👉 https://clawhub.ai/uncleJim21/pullthatupjamie
This is not another summarizer.
It is a visual, interactive podcast intelligence layer for AI agents. Search returns timestamped audio clips that can be assembled into fully navigable research sessions. Playable. Spatial. Grounded in real sound.
If your agent produces text walls, you are leaving signal on the table.
Visual, Human-First IntelligenceVisual, Human-First Intelligence
Most AI tools return paragraphs.
PullThatUpJamie returns artifacts.
For example:
https://www.pullthatupjamie.ai/app?sharedSession=c0bed41d939a
<div style="aspect-ratio: 1 / 1; border-radius: 12px; overflow: hidden; border: 1px solid rgba(255, 255, 255, 0.1); background-color: rgb(0, 0, 0); box-shadow: rgba(0, 0, 0, 0.5) 0px 4px 32px; position: relative;"><iframe src="https://www.pullthatupjamie.ai/app?sharedSession=c0bed41d939a&embed=true" title="Jamie — Explore Ideas" allow="autoplay" style="width: 100%; height: 100%; border: none;"></iframe></div>
That link is fully interactive. You can:
- Click between clips
- Play exact moments in audio
- See how topics connect
- Explore relationships visually
It is not a PDF.
It is not a blob of generated prose.
It is an explorable knowledge object grounded in primary source media.
Agents generate the structure. Humans explore it.
That design choice matters.
The Corpus Behind ItThe Corpus Behind It
Under the hood, the ClawHub skill exposes semantic search across:
- 109 podcast feeds
- ~7,000 episodes
- ~1.9 million indexed transcript segments
Coverage includes Joe Rogan Experience, Lex Fridman, Huberman Lab, TFTC, All-In, No Agenda, What Bitcoin Did, Acquired, and 100+ more.
Long form. High signal. Already transcribed and indexed.
Search returns in under two seconds.
And it does not return summaries.
It returns:
- Timestamped audio clips
- Direct deeplinks
- Exact transcript quotes
- Structured metadata
You press play and verify.
Cost AsymmetryCost Asymmetry
If you tried to build even a fraction of this yourself, you would pay for transcription, embeddings, storage, and engineering time.
PullThatUpJamie costs $0.002 per search.
Five hundred sats covers an entire deep research session.
Instead of indexing ten episodes yourself, you can run hundreds of semantic queries across thousands.
This is infrastructure, not a feature.
Intelligence, Not Chat OutputIntelligence, Not Chat Output
With the ClawHub skill installed, agents can:
- Research what a specific guest said across shows
- Analyze how a company or topic is discussed over time
- Pull primary source clips for claims
- Compare narratives across feeds
- Build interactive research sessions
- Feed structured JSON into downstream workflows
The output can be:
- A single clip deeplink
- A fully interactive research session
- Raw API data
Again, the key difference is this: the deliverable is not just text. It is something you can explore, click, and audit.
That reduces hallucination risk and increases trust.
The Machine Payable Web in PracticeThe Machine Payable Web in Practice
Most APIs assume a human operator.
Sign up.
Generate API key.
Manage billing in a dashboard.
That is SaaS era thinking.
PullThatUpJamie supports Lightning.
Request invoice.
Pay invoice.
Receive preimage.
Use preimage as authorization.
Your credential is proof of payment.
Machine to machine settlement.
There is a generous free tier so agents can experiment without friction. But the architecture assumes something bigger: agents with wallets, services priced per action, programmable commerce between tools.
That is the Machine Payable Web in action.
Not theory. Running code.
Alby Hub and the Alby CLI make this possible. Instead of pasting API keys into dashboards, agents can use Nostr Wallet Connect to operate with a sandboxed Lightning wallet. One connection string gives your agent a budget and the ability to pay invoices programmatically. Request invoice. Pay. Receive preimage. Use it as authorization. That flow turns Lightning into a native machine credential. It is simple, auditable, and built for agents instead of humans.
If you are building on ClawHub and want verifiable, interactive podcast intelligence instead of synthetic summaries:
Install the skill:
https://clawhub.ai/uncleJim21/pullthatupjamie
Agents do not need more words.
They need ground truth you can press play on.
Whoa, this looks next-level. Not just summaries, but actual timestamped clips you can explore? That’s insane for research, feels like you can really audit everything instead of trusting a bot. Curious how fast it actually is across thousands of episodes.