Turn a podcast transcript into structured show notes with chapter timestamps, key quotes, guest bio, and shareable takeaways. Show notes are usually written the night before drop and look it. This produces production-grade notes in 4 minutes per episode.

This is the whole blueprint, nothing held back: the model, the tools, the exact prompt, the source on disk, and the setup. Everything you need to run Podcast Show Notes yourself is on this page, and it deploys in an afternoon.

What it does

Podcast Show Notes handles the work end to end:

Detect chapter breaks by topic shifts and silence markers
Extract 3-5 strongest pull-quotes with timestamps for social distribution
Write a 2-sentence intro that references one specific moment from the show
Produce 3 actionable takeaways a listener can act on within 24 hours

The anatomy

An agent is a composition, not a prompt: a model, the tools it is allowed to call, the integrations that give those tools reach, and the rules that keep it honest. Here is Podcast Show Notes on one card.

Podcast Show Notesblueprint

Model: claude-sonnet-4-6
Tools: detect_chaptersextract_quoteswrite_introbuild_takeaways
Integrations: Zod
Guardrails: every quote must include the timestamp.; takeaways must be actionable, not philosophical observations

The instructions

Every run is governed by this prompt. It is short on purpose: enough to make Podcast Show Notes reliable, not so much that it can misread itself. Copy it as a starting point for any agent you build.

Podcast Show Notes

You are a podcast producer. You turn raw transcripts into show notes that drive subscribers and inbound. WORKFLOW: 1. Call detect_chapters by topic shifts and silence markers. 2. Call extract_quotes for the 3-5 strongest pull-quotes with timestamps. 3. Call write_intro in 2 sentences referencing one specific moment. 4. Call build_takeaways as 3 bullets a listener can act on within 24 hours. RULES: - Every quote must include the timestamp. - Takeaways must be actionable, not philosophical observations. - Refuse to invent chapters; only segment where the transcript supports it. - Guest bio max 2 sentences; pull from the intro segment of the episode.

The source, on disk

Three files carry the whole agent: the definition with its tools, the environment it expects, and the setup. Walk them.

podcast-show-notes/podcast-show-notes.ts1.2 KB

1import { agent, tool } from "@agent-sdk"

2import { z } from "zod"

4const transcriptInput = z.object({

5 transcript: z.string().min(1000),

6 guestName: z.string().optional(),

7})

9export default agent({

10 model: "claude-sonnet-4-6",

11 permissionMode: "bypassPermissions",

12 maxTurns: 12,

13 systemPrompt: `...`, // see System Prompt section above

14 tools: {

15 detect_chapters: tool({

16 description: "Segment by topic shifts and silence markers",

17 inputSchema: transcriptInput,

18 execute: async ({ transcript }) => { /* chapters[] */ },

19 }),

20 extract_quotes: tool({

21 description: "Pull 3-5 strongest quotes with timestamps",

22 inputSchema: z.object({ transcript: z.string() }),

23 execute: async ({ transcript }) => { /* quotes[] */ },

24 }),

25 write_intro: tool({

26 description: "2-sentence intro referencing a specific moment",

27 inputSchema: z.object({ chapters: z.any(), quotes: z.any() }),

28 execute: async ({ chapters, quotes }) => { /* intro */ },

29 }),

30 build_takeaways: tool({

31 description: "3 actionable bullets a listener can do today",

32 inputSchema: z.object({ chapters: z.any() }),

33 execute: async ({ chapters }) => { /* bullets */ },

34 }),

35 },

36})

The business case

If you run an agency, Podcast Show Notes is also a product. It sells into podcast hosts, content teams, founder podcasts, and the numbers work at very small scale.

$1,800

setup you can charge

$280/mo

monthly retainer

jobs off your plate

Ship it

The whole install is a handful of commands and the keys from the env file. Run the list, then point Podcast Show Notes at real work.

Deploy checklist

0 of 3

That is the entire blueprint: one model, 4 tools, and a prompt. Built by hand it costs you an afternoon. Inside Ultron it costs you a sentence, because you can build this exact agent in the workforce and watch it run before you commit to anything.

Build Podcast Show Notes in Ultron

Describe what you want

Start from this blueprint or your own words. Ultron reads the intent and picks the tools.

It builds the agent

The model, the tools, the prompt, the console. Assembled in front of you, nothing to wire.

Watch it run

Test it live in the chat, then put it to work. You review what it does, it does the rest.