Skip to main content
Eye on AI

#313 Jonathan Wall: AI Agents Are Reshaping the Future of Compute Infrastructure

52 min episode · 2 min read
·
Jonathan Wall

Episode

52 min

Read time

2 min

Topics

Remote Work, Startups, Fundraising & VC

AI-Generated Summary

Key Takeaways

  • Agent Compute Primitives: Agents need isolated virtual machines rather than traditional server infrastructure because they exhibit unpredictable resource usage, write their own code, download documents dynamically, and require access to full computing environments including bash terminals and file systems. RunLoop's dev box provides each agent its own containerized micro-VM with complete tool access while maintaining security boundaries through isolation.
  • Benchmarking for Accuracy: RunLoop's benchmarking system lets developers create domain-specific tests with known starting states and desired outcomes, enabling rapid iteration on agent performance. Companies can test model changes (like switching from one LLM to another), prompt modifications, or framework updates by running agents against consistent benchmarks and measuring score improvements, reducing reinforcement learning cycles from weeks to hours.
  • Agent Deployment Workflow: Developers build agents locally using frameworks like Claude Agent SDK, Langchain DeepAgents, or Codex SDK, then deploy to RunLoop via API. When end users trigger actions (GitHub pull requests, Zendesk tickets, Slack mentions), the system spins up isolated dev boxes, mounts necessary context, executes the agent with full computer access, and tears down the environment upon completion.
  • Enterprise Adoption Pattern: Most effective agent implementations follow an 80-20 model where agents handle the bulk of workflow execution while humans audit and approve final results. This mirrors existing code review patterns where one engineer writes code and another reviews before merging. Companies should start with benchmarking and latest models before considering supervised fine-tuning for high-volume use cases or reinforcement learning for business-critical applications.
  • Workforce Integration Reality: Engineering teams already operate with multiple agents per person, with individual developers choosing different tools (Gemini, Claude, custom agents) based on personal workflow preferences and specific tasks. Rather than top-down corporate mandates, agent adoption follows individual experimentation patterns where coworkers with similar roles use different agents in different ways, similar to how people have distinct research and writing processes.

What It Covers

Jonathan Wall, founder of RunLoop AI, explains how AI agents require fundamentally different compute infrastructure than traditional servers. RunLoop provides isolated virtual machine environments (dev boxes) where agents can safely execute unpredictable workloads, access tools, and operate with their own dedicated computing resources, enabling companies to deploy and benchmark thousands of agents simultaneously.

Key Questions Answered

  • Agent Compute Primitives: Agents need isolated virtual machines rather than traditional server infrastructure because they exhibit unpredictable resource usage, write their own code, download documents dynamically, and require access to full computing environments including bash terminals and file systems. RunLoop's dev box provides each agent its own containerized micro-VM with complete tool access while maintaining security boundaries through isolation.
  • Benchmarking for Accuracy: RunLoop's benchmarking system lets developers create domain-specific tests with known starting states and desired outcomes, enabling rapid iteration on agent performance. Companies can test model changes (like switching from one LLM to another), prompt modifications, or framework updates by running agents against consistent benchmarks and measuring score improvements, reducing reinforcement learning cycles from weeks to hours.
  • Agent Deployment Workflow: Developers build agents locally using frameworks like Claude Agent SDK, Langchain DeepAgents, or Codex SDK, then deploy to RunLoop via API. When end users trigger actions (GitHub pull requests, Zendesk tickets, Slack mentions), the system spins up isolated dev boxes, mounts necessary context, executes the agent with full computer access, and tears down the environment upon completion.
  • Enterprise Adoption Pattern: Most effective agent implementations follow an 80-20 model where agents handle the bulk of workflow execution while humans audit and approve final results. This mirrors existing code review patterns where one engineer writes code and another reviews before merging. Companies should start with benchmarking and latest models before considering supervised fine-tuning for high-volume use cases or reinforcement learning for business-critical applications.
  • Workforce Integration Reality: Engineering teams already operate with multiple agents per person, with individual developers choosing different tools (Gemini, Claude, custom agents) based on personal workflow preferences and specific tasks. Rather than top-down corporate mandates, agent adoption follows individual experimentation patterns where coworkers with similar roles use different agents in different ways, similar to how people have distinct research and writing processes.

Notable Moment

Wall reveals that in just 18 months, coding agents evolved from basic ChatGPT copy-paste assistance to systems capable of writing 60-90% of production code autonomously. Depending on complexity, agents can now build complete applications with databases, authentication, and user interfaces without human intervention, though developers still direct, review, and occasionally restart agents when quality issues arise.

Know someone who'd find this useful?

Episode Transcript

I thought there's it would be interesting to apply the lens of infrastructure to kind of the coming agentic revolution. Got together with some former teammates from my previous startup and hired some new folks to start run loop. Right. And so the agent operates in its own virtual machine, not necessarily a web page, not necessarily a browser. Or is it a browser? Agents work differently enough and have different enough compute patterns that they need their own new compute primitive, and that is core to what we're building here at Unloop. We call that our dev box. So there are some folks that that make agent builders, and that is super useful and super necessary. Some of them are paired to different frameworks, Like, the folks at Langchain have a cool agent builder that is is tightly coupled to to Langchain, for example. But where we're playing is a little lower. It's a little closer to the infrastructure layer. And we ultimately wanna be the run time where the agent is deployed and where it executes. So what I generally do is have you introduce yourself, give your background as far as it's relevant, and how you got to run loop, and then talk about what run loop is. And then we'll talk. There's some of the stuff in the materials that was sent over that was interesting to me. You talk about agents changing the shape of compute. Yeah. And I'd like to hear all of that. But let's start with the introduction. Okay. Yeah. Hello. It is it's great to meet you, Craig. My name is Jonathan Wall. I am the founder of RunLoop AI, I suppose, the CEO as well. I cut my teeth early in my career at Google. I joined Google not long after it it went public where I was the tech lead of the Google file system. So that was the seminal early days of huge, scalable distributed systems. So I had a front row seat and seeing, like, kind of an explosion of the Internet being realized and new technology being built for it. Around 2009, another engineer and I went out and founded the Google Wallet. So that was the first NFC solution for tap and pay. When you see people tapping phones to pay, that was the technology we created in 2009. And at the time, I didn't think about it in as much detail, but it was picking up on the next big platform revolution in technology at the time, which was the mobile phone and the kind of the new capabilities that FormFactor enabled. Went from the build out of the Internet in at scale infrastructure to working on payments in mobile. From there, I left and started a company with another gentleman from the Google Wallet effort called Index, and we did secure payments infrastructure for tier one retailers. And that company was ultimately acquired. It was named Index. It was ultimately acquired by Stripe …

Get the full transcript (8,705 words) + summary by email — free

One-time email with the complete transcript and AI summary of this episode. No account needed.

One email, no spam. We’ll also show you what SignalCast does.

Browse all Eye on AI transcripts →

You just read a 3-minute summary of a 49-minute episode.

Get Eye on AI summarized like this every Monday — plus up to 2 more podcasts, free.

Pick Your Podcasts — Free

Keep Reading

Books, tools, and gear mentioned in this episode

SignalCast may earn commission on purchases via these links.

Tools

  • Developers build agents locally using frameworks like Claude Agent SDK, Langchain DeepAgents, or Codex SDK, then deploy to RunLoop via API.
  • Developers build agents locally using frameworks like Claude Agent SDK, Langchain DeepAgents, or Codex SDK, then deploy to RunLoop via API.
  • by Anthropic

    Developers build agents locally using frameworks like Claude Agent SDK, Langchain DeepAgents, or Codex SDK, then deploy to RunLoop via API.

company

  • RunLoop AIBy guest
    Jonathan Wall, founder of RunLoop AI, explains how AI agents require fundamentally different compute infrastructure than traditional servers. RunLoop provides isolated virtual machine environments (dev boxes) where agents can safely execute unpredictable workloads.

More from Eye on AI

We summarize every new episode. Want them in your inbox?

Similar Episodes

Related episodes from other podcasts

Explore Related Topics

This podcast is featured in Best AI Podcasts (2026) — ranked and reviewed with AI summaries.

Read this week's Startups & Product Podcast Insights — cross-podcast analysis updated weekly.

You're clearly into Eye on AI.

Every Monday, we deliver AI summaries of the latest episodes from Eye on AI and 192+ other podcasts. Free for one show.

Start My Monday Digest

No credit card · Unsubscribe anytime