Teardown No. 32 · Outbound

How Yoni Argaman built three Claude Cowork agents that find and pitch distribution channels

Yoni Argaman was VP Marketing at Nilus when he wrote this post. He wanted to reach CFOs and Treasurers through conferences, podcasts, communities, influencers and media. He wanted to do it systematically, without adding headcount. He built three agents on Claude Cowork. A Discovery Agent runs every morning and sends new channels to his Slack DM. An Outreach Agent researches a chosen channel and drafts a tailored pitch. A Pipeline Report arrives every Monday. He approves each draft with a Slack emoji reaction. In the first week the Discovery Agent surfaced 53 channels he had not formally tracked. One Outreach run produced drafts for three channels. He states no replies or bookings.

Yoni ArgamanVerified Oct 17, 2026Markdown copy

The problem

He describes distribution channel work as “grinding.” The work means searching for podcasts, newsletters, communities and events. It also means finding the right contact, writing a pitch tailored to the audience, and tracking what has been sent, what is pending and what has been ignored. He says most teams end up with “a spreadsheet that’s 60% outdated.” He set a goal to “put Nilus in front of at least 75% of our addressable personas through the right distribution channels.” He wanted to do it “not with more headcount” and “systematically rather than reactively.”

The blueprint (as he describes it)

  1. Define the audience. His target persona is CFOs and Treasurers at growth-stage and mid-market companies in North America. He calls it a specific audience that moves through a defined set of channels.
  2. Discovery Agent, every morning. It searches for new channels that target CFO and Treasury personas. It covers conferences, podcasts, online communities, LinkedIn influencers and media publications. It deduplicates against everything it has already found. Then it delivers a formatted briefing to his Slack DM.
  3. Outreach Agent, on demand. To activate a channel, he mentions its name in Slack and triggers the agent. The agent finds the right contact person, reads their recent content and identifies why Nilus is a fit. Then it drafts a personalized partnership pitch. He says it is “Not a template.” The drafts are “ready to copy and send.”
  4. Pipeline Report, every Monday morning. It reads the week’s activity. It counts what was discovered, drafted, approved and sent. It flags anything sitting idle. It also adds one observation about what is working or missing in the pipeline.
  5. Approval in Slack. He reacts with emojis directly on each draft (✅ approve / ❌ skip). His advice is “Build your approval workflow first.” He says the bottleneck shifts to the human, who reviews, approves and sends.
  6. Pick the output layer that works. He says “Integrations are the bottleneck.” Writing back to a Google Sheet was blocked by an egress proxy. Creating Gmail drafts from scheduled tasks failed silently. Posting to a shared Slack channel failed because the Slack app was not a member of the channel. The reliable path was a Notion doc or a private Slack DM.

Stated results

  • First week of discovery: 53 channels he “hadn’t formally tracked.” His examples:

    • The AFP FP&A Forum (500+ CFOs, 12 days away at the time).
    • CFO Brew (100,000+ subscribers).
    • Jack Sweeney’s CFO THOUGHT LEADER podcast (600+ CFO interviews).

    These audience figures describe the channels. They are not campaign results.

  • One Outreach run produced drafts for three channels:

    • A sponsorship pitch to Jeff Schiz at AFP.
    • A sponsored content proposal to CFO Brew’s editor that references their five recent AI-in-finance pieces.
    • A guest podcast pitch to Jack Sweeney for his Planning Aces series.

  • Replies, sponsorships, bookings or coverage: None stated.

  • Progress toward the 75% persona-reach goal: None stated.

Caveats

  • This is a personal Substack post. It does not sell a product, course or kit. The author was VP Marketing at Nilus, and the post describes Nilus and its backer Bessemer, so it partly promotes his then-employer.
  • He was a marketer at a venture-backed company with a team (“We tried…”), not a solo operator.
  • The results are counts of channels found and pitches drafted. He does not say whether any of the three drafts were sent or answered.
  • He does not describe the agent prompts, the sources the Discovery Agent searches, or how the Notion doc fits with the Slack DM.
  • His claims that continuous discovery compounds and that automation raises activation rate are his opinion, with no figures.
  • His LinkedIn now lists Newsbound AI, so he may have left Nilus since the post.
  • Handles: Substack @yoni1492 (verified 2026-10-17: author metadata on his own Substack post). LinkedIn linkedin.com/in/yoniargaman (verified 2026-10-17 by name plus NYU Stern match with his Substack bio). X: not checked, no handle used.

Sources

Want agents like these working your distribution? See the Agent Stack Teardown. Get new teardowns weekly: subscribe (double opt-in).

Get one teardown a week.

Same format, a new operator every issue. Double opt-in.