Cody Schneider (Graphed): What It Takes to Build AI Marketing Agents That Work
カートのアイテムが多すぎます
カートに追加できませんでした。
ウィッシュリストに追加できませんでした。
ほしい物リストの削除に失敗しました。
ポッドキャストのフォローに失敗しました
ポッドキャストのフォロー解除に失敗しました
-
ナレーター:
-
著者:
概要
Episode Stack: https://stackl.ist/4vWPzAl
You've heard the pitch: AI agents that run your marketing, write your content, manage your ads, source your leads. It sounds like science fiction, or at minimum, like something that works for someone else's company. Cody Schneider has actually built it. In this episode, Cody, co-founder of Graphed, former growth lead at Rupa Health, and serial builder from Coeur d'Alene, Idaho — sits down with Kyle to get specific about what's real and what's hype in the AI go-to-market space. He's not philosophizing from the outside. He built a sourcing agent before breakfast the morning of this recording.
Discussion topics:
- Why "just give an agent a task" fails, and what agent orchestration actually looks like when it works
- The AI SDR post-mortem: why those tools didn't fail because AI is bad, and what the wrong optimization metric actually costs you
- How Cody thinks about "biology": whether an idea has the structure to become a real company or a good side project
- The compounding go-to-market loop: layering channels, acting on signal, and why you never stop what's working
- Data quality as the hidden killer of every AI analytics project (including why your Facebook ads API data is probably wrong)
- Vibe coding from 0 to 80% vs. 80% to production & why that gap is where companies get stuck
- What a GTM engineer actually does in 2026, and why that skillset is one of the rarest in the market right now
- The future Cody is betting on: agent teams, services bought as outcomes, and what happens when the cost of intelligence approaches zero
Key moments:
[00:00] Board meetings with your agents
[01:00] An agent Cody built before 8am and what it did [04:12] How Cody evaluates whether an idea has "biology" to be a real company
[08:56] From Etsy scraping to Rupa Health: the origin story
[15:48] Signal and noise — acting fast when you have it
[18:00] Why AI SDRs failed (it wasn't the agents)
[22:12] Software that molds to the user: the end of dropdown UIs
[27:12] Vibe coding's dirty secret
[33:06] Where Cody sees Graphed in six months
[49:15] Cody's homework for non-technical founders right now
[52:30] Kyle's live Graphed demo: Apollo, Stripe, QuickBooks, Brex, and PostHog in one afternoon