Episode

E191: Super Fast Infra for Agents to Use the Internet

Podcast: Open Source Startup Podcast
Published: Feb 4, 2026
Duration seconds: 2173
Processing state: processed
Canonical source: https://podcasters.spotify.com/pod/show/ossstartuppodcast/episodes/E191-Super-Fast-Infra-for-Agents-to-Use-the-Internet-e3ejt3h
Audio: https://anchor.fm/s/3eab794c/podcast/play/114995761/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-1-4%2F7572ef78-ed5f-e4a4-d3e7-fe83f15afb34.mp3
JSON: /v1/public/podcasts/open-source-startup-podcast/episodes/e191-super-fast-infra-for-agents-to-use-the-internet
Markdown: /podcast/open-source-startup-podcast/e191-super-fast-infra-for-agents-to-use-the-internet.md

Actions

POST https://stenobird.com/v1/public/podcasts/open-source-startup-podcast/episodes/e191-super-fast-infra-for-agents-to-use-the-internet/transcription-requests
Idempotently request low-priority transcript generation for this episode.
GET https://stenobird.com/podcast/open-source-startup-podcast/e191-super-fast-infra-for-agents-to-use-the-internet.md
Read the agent-friendly Markdown representation of this episode resource.

Summary

Kernel provides high-performance, low-latency browser infrastructure designed specifically for AI agents to navigate the live web. The discussion explores how moving from heavy Docker containers to unikernels and micro-VMs enables scalable, real-time web automation.

Topics

AI Agents
Browser Infrastructure
Cloud Hypervisor
Micro-VMs
Web Automation
Open Source Software
RPA
Unikernels

Highlights

Main idea: Browser infrastructure is becoming a critical layer for AI agents to interact with the live internet
Technical shift: Moving from standard Docker containers to Cloud Hypervisor and micro-VMs allows for features like real-time memory hot-swapping
Practical takeaway: Developers can use Kernel's APIs or MCP servers to run large-scale browser automation without managing complex Kubernetes clusters
Use case: High-performance browsers enable new forms of RPA for websites without APIs, such as real-time price tracking and sales intelligence
Failure mode: Relying on heavy, high-latency infrastructure prevents agents from performing real-time tasks like voice-based web interaction

Chapters

1:00 Founding Story and AI Evolution: Catherine discusses her background in early AI waves and how her experience with supervised machine learning led to the creation of Kernel.
3:45 Identifying the Infrastructure Gap: The realization that scaling AI agents requires a specialized infrastructure layer to handle production-grade web automation.
6:30 From Docker to Unikernels: A technical deep dive into experimenting with unikernels and Firecracker VMs to achieve extreme performance.
9:10 Browser-as-a-Service Overview: Defining Kernel's core offering: providing scalable, low-latency browser access for autonomous agents.
11:55 Use Cases for Agentic Web Access: Exploring how agents use live web access for market landscape analysis, price tracking, and data set building.
14:35 Developer Accessibility and Tiers: How Kernel uses free and hobby tiers to lower the barrier to entry for developers building agentic workflows.
17:15 The Role of Open Source in Distribution: Discussing the technical trajectory of the control plane and using open source as a tool for community trust and distribution.
22:35 The Business Value of Low Latency: Why performance is a fundamental requirement for building viable consumer-facing agent products.