🎙️ New episode. What happens when the angriest person on your call list gets an AI on the line instead of a human?

Seb Hapte-Selassie, Stanford CS, ex-Pitch, and Co-founder & CTO of telli (just raised a €15M seed led by redalpine), joined Hanano and Joseph to talk about YC, hiring, and building AI people don't mind talking to.

Here's what we unpacked:

📞 An outbound call to someone still angry about an old issue, who calmed down and booked an appointment. Seb's point: agents have infinite patience, and call centre work is exhausting in a way that makes that hard for humans.

🧠 What he learned from his mother, the most high-agency person he's known: most barriers you assume are there simply aren't.

🧩 Why a big lab shipping a product doesn't mean it wins the category. Model providers do well when you burn more tokens. Application companies do well when they find the cheapest model that solves the problem, which is an argument for staying model-agnostic.

🤖 Why he thinks agents belong inside Slack, Teams, HubSpot and Salesforce rather than in yet another dashboard. Their new agent Charlie works that way, and it says something about where software is heading.

⚙️ Multiple models running in parallel on a single call, with fallbacks when a provider fails. They'll even switch models mid-conversation depending on the task.

🚫 Which customers do you say no to? No cold calling, no gambling, nothing that tricks people.

🎓 His advice for a CDTM student who wants to start an AI agent company. It isn't "start now."

Tune in now: cdtm.me/seb

Originally posted on LinkedIn · View the original post