superagnt_
// offer letter // teammate
ready to start immediately

YouTube Partnership Channel Scraper — every-3-days discovery of partnership channels

Every 3 days, find YouTube channels making content adjacent to your product — filter by audience size + country + ICP relevance, fetch contact emails, write to a deduped partnership table. Pair with the YouTube Outreach Writer to actually pitch them.

YouTube
Claim this hireno questions · deploys on claim
01// what they own

The outcomes on their scorecard

Tier-classified partnership candidates

Each channel gets tagged small / mid / large based on subscriber count. The outreach writer uses the tier to pick a different voice + offer per group.

ICP-judged on the actual content, not the keywords

The LLM relevance gate runs on recent video titles + channel description — what the channel actually makes — not on whether they showed up in a keyword search.

Per-row commits, crash-safe

Each new channel is committed as soon as enrichment finishes. A mid-run crash doesn't lose the channels that already landed.

Email-gated

Channels without a discoverable email are dropped — unreachable channels never clutter the table.

02// the motion

How the work gets done

YouTube creators with audiences that overlap your ICP are partnership goldmines — sponsorships, integrations, mutual promotion, affiliate. But finding them by hand is brutal: searching YouTube, sifting through hundreds of channels per topic, manually checking which ones have contact emails. This agent automates the discovery. Every 3 days it runs a TypeScript skill that searches YouTube for your topic phrases, dedupes against what's already discovered, fetches subscriber counts + country + recent videos for each new candidate, applies cheap filters first (audience size, country), then LLM-judges ICP relevance from the channel's actual recent video titles. Survivors get email-fetched; survivors-with-emails get a one-paragraph channel summary written by the LLM and saved with a tier classification (small / mid / large). Per-row commits — if the skill crashes mid-run, what's saved stays saved. Pair with `youtube-partnership-outreach-writer` for the activation half.

Every 3 days, the agent searches YouTube for your topics, dedupes, fetches subs + country + recent videos for each new candidate, LLM-judges ICP relevance on the actual channel content, fetches emails for survivors, and writes everything to a partnership table. Per-row commits keep the run resilient.

click a node to inspect
03// the roster

Your new teammate

01agent

YouTube Partnership Channel Scraper

Every-3-days scheduled agent that searches YouTube for channels in my space, filters by audience size + country + ICP relevance, fetches contact emails, and writes new channels to youtube_partnership_channels. Per-row commits. Pairs upstream-of the YouTube Outreach Writer.

1 schedule
04// built on

What it runs on

This hire is assembled from the public toolkit. Every piece below is a tool your own agents can call too.

05// onboarding

How setup works

  1. 01

    Claim the hire

    One click. No questionnaire. The brief is already written from your business, and you can edit it before you build if you want to.

  2. 02

    Your hire deploys itself

    Agents, tools, schedules, memory, hand-off queues and the database schema are provisioned into your workspace automatically, then the first real run fires while you watch.

  3. 03

    Hand it the keys it still needs

    You land in chat with your hire and its findings. 1 setup tasks — connecting your outreach tools, picking where it reports — are done conversationally, whenever you want.

06// related

Other hires that pair well

» build it yourself

Every hire is assembled from the same toolkit.

Same data sources, same database, same memory, schedules and queues. Point your own agent at the MCP endpoint and wire it your way instead.

Wire your own agent