AI + a16z Podcast: Automating Developer Email with MCP and AI Agents

Date: March 21, 2025
Guests: Yoko Lee (a16z Partner), Zen Orocha (Resend Founder & CEO), Derek (Host)

Episode Overview

This episode explores the rapid shift in developer experience as AI agents increasingly automate tasks traditionally performed by humans, especially in domains like email. Zen Orocha, founder of Resend, and a16z partner Yoko Lee discuss the implications of agent-centric workflows, the evolution of tooling and standards like MCP, and how generative AI is democratizing the creation of highly tailored developer tools and email experiences. The discussion is sprinkled with stories – from the practical (streamlining onboarding for bots) to the personal (the wild origin story of the wildly popular Dracula code theme).

Key Discussion Points & Insights

1. The Rise of Agent-Centric Development

• Shift from Human to Agent Actions:

  • AI agents are increasingly becoming first-class actors in systems where humans previously dominated, forcing products to be rethought for machine interactions.
  • "I believe it's going to be the majority of the actions will be taken by agents instead of humans. And that's just the reality we're going to live in." — Zen Orocha [00:00, 41:12]

• Redefinition of Developer (and Agent) Experience:

  • Developer experience (DX) must now account for agent experience—products, onboarding, and permissions need to accommodate both.
  • "As a SaaS provider ourselves, we are thinking about how we can make the product easier for agents to consume." — Zen [02:16]
  • "In our industry, you go to SendGrid... manual verification... two days for you to get approval. In an agent first world, that's simply not possible." — Zen [03:34]

2. Email in the AI and Agent Context

• From Pretty HTML to Plaintext for Agents:

  • While humans respond to visually rich HTML emails, agents process structured plain text more effectively (fewer tokens, easier parsing).
  • "Turns out in today's world, that's actually better than the HTML version because it's easier to parse, you're using less tokens, it's cheaper." — Zen [06:43]
  • "The more content focused these emails are, the less markup... if they have a plain text version of it, I think this is more important than ever." — Zen [07:45]

• Agents as Email Users:

  • Agents could own their own emails, send, receive, and parse content, creating a new class of automated workflows.

3. AI Tools Unlocking Creativity and Speed

• Democratizing Creation:

  • LLM-powered tools put powerful capabilities in the hands of technical and non-technical users, rapidly collapsing creative loops for things like email and UI prototypes.
  • Example: "New Email," a Resend tool, lets users generate email templates in seconds—even non-developers.
  • "Before you needed to maybe hire an agency... now we can give that to LLMs and empower the actual creator." — Zen [14:00]

• Prosumer AI Apps:

  • A wave of specialized, vertical AI apps (e.g., V0, Lovable, Bolt) let users rapidly build and deploy products, each with unique "aha" moments (like publishing a site or sending an email).
  • "But what really matters is what's the thing on the top right corner... it's publishing, it's going live... I think you know there's going to be apps for each one and that's the main part of each one." — Zen [08:47]

4. MCP: The Emerging Agent/Tool Protocol

• What is MCP?

  • Machine Control Protocol (MCP) is becoming the connective tissue between agents, APIs, and services—especially if adopted by major AI model providers like OpenAI.
  • "The emerging standard is definitely MCP... Imagine if OpenAI adopts that, then it's over." — Zen [21:16]

• Ecosystem, Clients, and Servers:

  • Services like Resend aim to be both MCP servers and clients, supporting agent-driven workflows across products.
  • "I can definitely see a world where we are the MCP client where you can come in and say... get my top 10 linear feature requests and then draft an email." — Zen [23:34]

• Challenges/Opportunities:

  • Adoption by major players and support for multiple interfaces needed for ecosystem growth.
  • "It's an ever-evolving ecosystem, right? There's no right answer yet." — Zen [21:16]

5. State, Databases, and Agent Infrastructure

• Databases Behind Agent Workflows:
  • Rapid, frictionless database provisioning (often using tools like Supabase, Neon) is key to scaling agent-first apps, especially as agents need to persist state.
  • "The race... is going to be who can reduce friction the most who can spin up a new postgres database faster." — Zen [27:08]

6. Notable Use Cases and Stories

• AI-Powered Newsletters:

  • Recurring, personalized newsletters generated daily from multiple data sources through agent workflows.
  • "There are people with thousands of domains in one recent account and they're spinning up new domains as new applications are built." — Zen [29:55]

• Programmatic Domain Management and Prewarming:

  • Automated domain and SMTP setup for new apps, e.g., when provisioning payload CMS instances via Resend.
  • "Every time they provision one of those, they provision a new domain powered by Resend. And then email sending is already done for you." — Zen [30:55]

• Agents Creating and Registering Their Own Domains:

  • Speculated future where agents iterate on side projects, register domains, and launch new services autonomously.
  • "Maybe they will come up with like new project ideas and they'll be like, oh, let me see if the domain is free. Oh yeah, it is. Buy." — Zen [31:43]

7. Advice for AI Developers

• Start with the Use Case, Not the API:

  • Focus on real pain points or curiosities, not API completeness.
  • "Start from the use case. What are people actually doing with your product? ... You don't need full coverage of your API, just the most important things." — Zen [34:25]

• Embrace New Tools, Even if They’re Uncomfortable:

  • "I had to like just one day stop and be like, okay, let me look into this thing. And I think it's around, like changing your tool set in a way." — Zen [33:03]

8. The Dracula Theme Origin Story – Creativity Out of Adversity

• Zen’s accidental hospital hackathon:
  • Zen, while hospitalized abroad, had his laptop stolen, received a new one from coworkers, and began building the Dracula theme to unify his coding environments.
  • "I just wanted a theme that worked everywhere. So I built the first version in the hospital and then it just took off after that." — Zen [39:07]
  • Dracula has >9 million users and gave Zen confidence to found Resend.

Notable Quotes & Memorable Moments

• "Content remains the king. The difference now is that we're going to be living in a world where agents are talking to other agents..." — Zen Orocha [07:45]
• "We can give that to LLMs now and empower the actual creator... back and forth between agencies and non technical users... it's just like wasted time." — Zen Orocha [14:00]
• "If OpenAI adopts [MCP], then it's over." — Zen Orocha [21:16]
• "You need to store state, right? ... Maybe we'll need a new kind of database." — Zen Orocha [27:01]
• "I hated the fact that I had a theme on my code editor and a different theme on my browser... so let me just build one that works everywhere." — Zen Orocha [35:52, 39:07]

Timestamps for Major Segments

• [00:00] The agent revolution in software: from human to agent actors
• [02:16] What is "agent experience" and DX for AI agents?
• [03:34] Product workflows rethought for agents (onboarding, verification, docs)
• [05:39] Email architecture: agents as users, plain text vs HTML, abstraction needs
• [08:47] Prosumer AI apps and generative creation patterns
• [14:00] LLMs democratizing creation—reducing friction for marketers/designers
• [21:16] MCP as the standard for agent-to-API interaction
• [27:08] Databases and state in agent workflows
• [29:55] AI-generated newsletters and automated domains
• [33:03] Career/Toolchain advice for entering AI dev
• [39:07] The Dracula theme origin story
• [41:12] Future of email, automation, and agent-first product design
• [42:53] Rethinking audit trails & accountability for agents

The Future Outlook

Zen and Yoko agree that as agents become the dominant actors in software systems, every layer—from onboarding to permission models and even the tracking of actions—will need to be re-envisioned for an agent-first world. Adoption of protocols like MCP and the leveraging of AI for creativity and productivity will not only save time but empower an entirely new (and larger) set of users. The transition may be messy, but the opportunity for both developers and end users is immense.

Final Takeaway

"You really have to go down the whole journey of the user... [as] products you have to rethink... Even the way you store activities... because here, humans are not going to be the only user anymore."
— Zen Orocha [41:57–43:14]

This episode offers both visionary perspective and actionable advice for developers and founders navigating the changing terrain of AI and agent-driven software.