I think there will be bad actors in any field, and right now, a lot of agent-based outreach might fall into that bucket, so its rational to be initially skeptical.

The more interesting shift isnt whether humans will keep using email with agents, but whether agents can become distinguishable from noise. Historically, we ignored anonymous calls but we engaged with known vendors that had reputation, contracts, and consequences.

Once an agent has a persistent identity/a domain, trust becomes something that can be accumulated over time instead of being assumed per message.