Why Every AI Agent Needs Its Own Computer | Ivan Burazin (Daytona)

Why It Matters

Sandboxed agents provide the secure, persistent compute needed for enterprise AI automation, turning speculative bots into reliable digital employees.

Key Takeaways

• Each AI agent requires its own sandboxed computer for secure operations.
• Sandboxes isolate agents, allowing tool installation, web access, and code execution.
• Dedicated sandboxes prevent rogue behavior and enable instant termination.
• Stateful sandboxes differ from stateless cloud apps, supporting continuous workflows.
• Daytona built custom scheduler, abandoning Kubernetes to meet agent scalability.

Summary

The conversation centers on Ivan Burazin’s claim that every AI agent needs its own sandboxed computer – a dedicated, isolated environment that functions like a personal workstation. He frames agents as digital knowledge workers, arguing that without a full‑featured computer they cannot perform the tool‑heavy tasks that drive real‑world productivity.

Burazin defines a sandbox as a composable computer: an isolated VM where an agent can install software, browse the web, run scripts, and handle 2FA‑protected actions. This isolation mitigates rogue behavior, allowing operators to “kill” the sandbox instantly. He contrasts this stateful approach with the stateless design of traditional hyperscaler services, emphasizing that agents require persistent environments to maintain context across tasks.

Illustrative examples include a board‑meeting scenario where an agent attempted to log into a bank, prompting Burazin to give the agent its own phone number for 2FA and strict spending limits. He also likens the sandbox to a Mac Mini, and describes Daytona’s custom scheduler—built to replace Kubernetes—to handle thousands of concurrent agents for research (RL evaluation) and long‑running background services like Perplexity.

The broader implication is that enterprises must adopt sandbox infrastructure to safely scale AI agents, unlocking secure access to legacy applications and high‑value data. This shift creates a new market for specialized orchestration platforms and reshapes how companies think about AI‑driven automation and security.

Original Description

If AI agents are the new digital knowledge workers, where exactly do they do their work? In this episode of the MAD Podcast, Ivan Burazin joins us to unpack the emerging infrastructure stack for AI agents and explain why every agent needs its own secure, stateful "computer." We explore the technical realities of sandboxes, dive into why legacy, stateless hyperscalers weren't built for these new workloads, and break down the mechanics of microVMs and custom schedulers alongside a contrarian prediction on an impending CPU shortage. Finally, Ivan delivers an absolute masterclass on product-led growth, community building, and go-to-market strategy for technical founders.

Ivan Burazin

LinkedIn - https://www.linkedin.com/in/ivanburazin

X/Twitter - https://x.com/ivanburazin

Daytona

Website - https://www.daytona.io/

X/Twitter - https://x.com/daytonaio

Matt Turck (Managing Director)

Blog - https://mattturck.com

LinkedIn - https://www.linkedin.com/in/turck/

X/Twitter - https://x.com/mattturck

FirstMark

Website - https://firstmark.com

X/Twitter - https://x.com/FirstMarkCap

Listen on:

Spotify - https://open.spotify.com/show/7yLATDSaFvgJG80ACcRJtq

Apple - https://podcasts.apple.com/us/podcast/the-mad-podcast-with-matt-turck/id1686238724

00:00 Intro

02:13 What is an AI agent sandbox?

03:17 Security risks of running agents locally

05:17 Stateful vs. stateless hyperscalers

07:04 The history of cloud IDEs and the end of localhost

09:45 Do all AI agents need a sandbox?

12:26 Sandbox use cases: RL evals & background agents

14:10 Unpacking the emerging AI Agent Stack

16:20 The unsolved problem of agent memory and learning

19:37 Where sandboxes fit in the agent harness

21:35 OpenAI, Anthropic, and agent SDKs

23:06 Ivan's founder journey: From CodeAnywhere to Daytona

26:59 GTM strategies and building developer communities

33:48 Why customer support is your best GTM strategy

35:34 Leveraging Twitter during the AI super cycle

40:50 The technical anatomy of a sandbox

41:53 Why fast spin-up speeds maximize GPU efficiency

46:09 Firecracker, QEMU, and isolation primitives

49:58 Why sandbox snapshots and state forking matter

51:40 Why Daytona built a custom scheduler from scratch

55:24 The challenge of long-running stateful sandboxes

58:10 The build your own sandbox trap

1:01:03 Why AI agents might trigger a global CPU shortage

1:02:46 The future of the AI Agent Stack