Build / Lovable · podcasts / audio (all shows - business, sports, tech, finance, academia)

Podcast Wire

A browsable, filterable feed of what the podcasts said this week - the readable takeaways AND the structured claims behind them, every one linked to the episode. A live, self-updating public site where anyone can browse this week's podcasts, read the takeaways, expand the structured claims, filter by show/topic/person/company, and jump straight to the real episode - all cited, none invented.

A public website that shows the past week of podcast episodes as browsable cards - each with a human-readable Brief (the narrative summary + key insight) and an expandable list of Signals (the atomic claims Synorb extracted), plus combinable filters (podcast, topic, entity, significance) and a link to every original episode.

Synorb extracts structured, cited claims from HOURS of spoken audio (210 podcast episodes in the last 7 days; ~48 claims/episode) - something no news API or web scraper can do. Podcasts are Synorb's most non-obvious superpower, and this turns them into a browsable, filterable intelligence product that reads like a premium publication and stays fresh by itself. It shows off both Synorb primitives at once: the human Brief for readers and the machine Signals for the curious/technical - side by side, each cited.

Click, answer a few plain-English questions, get a working product.

No setup to learn. Lovable asks a handful of normal-person questions — each with a smart default, so “you pick” always works — then builds it with live Synorb data flowing in.

1

Which podcasts should the feed cover?

Smart default: All podcasts (the broadest view - business, sports, tech, finance, academia), with quick-filter buttons; you can narrow to specific shows or topics after.

2

How should each episode show up - readable takeaways first, the structured claims first, or both side by side?

Smart default: Readable Brief up top with an 'expand claims' toggle that reveals the structured Signals underneath.

3

Pick a look: Editorial (magazine, serif, warm), Terminal (dark, dense, data-room), or Minimal (clean, lots of whitespace).

Smart default: Editorial - it reads like a premium publication and screenshots beautifully.

4

How fresh should it stay - refresh once a day or once a week?

Smart default: Once a day at 7am, covering a rolling 7-day window, so it's fresh every morning.

Pick a look. It ships with taste by default.

Editorial

Warm magazine aesthetic: serif headlines, generous spacing, the Brief's key_insight pulled out as a standfirst, show name + episode date as a byline, a subtle 'expand claims' toggle. Reads like a premium podcast-intelligence publication.

Terminal

Dark, dense data-room feel: monospace accents, compact episode rows, color-coded show + topic chips, significance dots, claims shown as a tight structured list. For an analytical/insider vibe.

Minimal

Clean and quiet: lots of whitespace, one accent color, each card foregrounds the key takeaway with a subtle 'Signals' expander and a 'Listen' chip. Distraction-free browsing.

Synorb is the cited core. Lovable’s own tools make it pop.

INFRA-BACKED, deployed Vite/React app on Lovable Cloud. BUILD-TIME: connect the Synorb MCP as a Lovable chat connector so the agent inspects real audio manifests (media_format=audio) and builds the Brief+Signals UI against the true shape. RUNTIME: a server-side function (an Edge Function, whatever this project uses) calls Synorb's REST API for the last 7 days of audio manifests (media_format=audio) plus any active filters (tag_names for topic/entity, source channel for the show), reading the Synorb API key + secret from server secrets - never in client code. A Supabase Cron job refreshes on the chosen schedule and upserts into Postgres so the site is fast. Credentials: Synorb is a real API and is THE data source - do not substitute another provider; I'll provide the key or you can ask me; Synorb Connect returns api_key, api_secret, and mcp_token.

Cited data core
Synorb
StreamsAll podcast source channels (media_format=audio) - business, sports (e.g. Locked On team pods), tech, finance, academia (e.g. Stanford GSB), ~210 audio manifests / 7 days, topic + person + org + place tags on each episode, home_domains span economics-business-work, arts-culture-entertainment, engineering-technology, and more
PrimitivesBrief (human-readable narrative + key_insight), Signals (atomic structured claims per episode), Tags (topic / person / organization / place) for filtering, source channel = the podcast/show, for the show filter, media_format=audio (the broad, correct way to get ALL podcasts), provenance citations (real episode source_url, never invented), significance level, coverage.status honesty
Experience layer
Lovable native tools
ComposesA browse/filter UI (combinable filters for podcast, topic, entity, significance - options populated from the week's real data), Recharts - a small 'most-active podcasts this week' or topic-mix chart from the real data, Native AI image generation - a themed hero/section art so it reads like a publication, A 'listen to the episode' link built from the real source_url (optional inline audio player), Supabase Cron - daily/weekly refresh so the feed stays current, Perplexity App Connector - GAP-FILLER ONLY (e.g. a show's site link Synorb doesn't carry), badged 'from the web', never mixed into cited claims
Web searchgaps only — clearly marked, never the primary feed

Credential safetySynorb credentials stay server-side or local — never in client/frontend code. You only enter them when the agent asks, with credentials from synorb.com/connect.

One click from the original source.

Provenance is the product. Every episode card carries a 'Listen to the episode' / source-domain link built from the real source_url Synorb returns, plus the show name (source channel) and episode date. Each expanded Signal claim and the Brief cite the same episode; a card with no source_url is suppressed rather than shown link-less. URLs are only ever the source_url Synorb returns - never invented.

Questions builders ask.

Will this show real live data on the first run?

Yes. The build prompt connects Synorb at build time to learn the real data shape, then wires live Synorb data into the running product. A live, self-updating public site where anyone can browse this week's podcasts, read the takeaways, expand the structured claims, filter by show/topic/person/company, and jump straight to the real episode - all cited, none invented.

Where do my Synorb credentials go?

Only where they are safe. INFRA-BACKED, deployed Vite/React app on Lovable Cloud. BUILD-TIME: connect the Synorb MCP as a Lovable chat connector so the agent inspects real audio manifests (media_format=audio) and builds the Brief+Signals UI against the true shape. RUNTIME: a server-side function (an Edge Function, whatever this project uses) calls Synorb's REST API for the last 7 days of audio manifests (media_format=audio) plus any active filters (tag_names for topic/entity, source channel for the show), reading the Synorb API key + secret from server secrets - never in client code. A Supabase Cron job refreshes on the chosen schedule and upserts into Postgres so the site is fast. Credentials: Synorb is a real API and is THE data source - do not substitute another provider; I'll provide the key or you can ask me; Synorb Connect returns api_key, api_secret, and mcp_token.

How are sources shown?

Provenance is the product. Every episode card carries a 'Listen to the episode' / source-domain link built from the real source_url Synorb returns, plus the show name (source channel) and episode date. Each expanded Signal claim and the Brief cite the same episode; a card with no source_url is suppressed rather than shown link-less. URLs are only ever the source_url Synorb returns - never invented.

Is this just a Synorb feed, or a real app?

A real app. Synorb is the cited data core; it also composes the platform's own tools: A browse/filter UI (combinable filters for podcast, topic, entity, significance - options populated from the week's real data), Recharts - a small 'most-active podcasts this week' or topic-mix chart from the real data, Native AI image generation - a themed hero/section art so it reads like a publication, A 'listen to the episode' link built from the real source_url (optional inline audio player), Supabase Cron - daily/weekly refresh so the feed stays current, Perplexity App Connector - GAP-FILLER ONLY (e.g. a show's site link Synorb doesn't carry), badged 'from the web', never mixed into cited claims.

What will it ask me before building?

A few plain-English questions, each with a smart default so you can just say 'you pick': Which podcasts should the feed cover?; How should each episode show up - readable takeaways first, the structured claims first, or both side by side?; Pick a look: Editorial (magazine, serif, warm), Terminal (dark, dense, data-room), or Minimal (clean, lots of whitespace).; How fresh should it stay - refresh once a day or once a week?.

Using a different agent?

The same Synorb MCP connects to every major coding agent. Pick yours.

Free credentials. No card.

Generate credentials, connect Lovable, and ship something with live, cited intelligence.