Oxide and Friends · 15 Jan 2026 · 1h 32m · Episode 2
Bryan Cantrill, Adam Leventhal, Rain Paharia, and David Crespo of Oxide Computer Company discuss using LLMs like Claude Code and Opus 4.5 to increase engineering rigor rather than just velocity. They share specific examples of using AI to refactor the Helios kernel, build the 'iddqd' Rust map library, debug the Ghostty terminal in Zig, and manage type drift in Oxide's APIs. The group argues that rigorous documentation and strong typing are essential for effective LLM collaboration, contrasting this with the 'vibe coding' narrative.
Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationSonnet 4.1less than a day (vs weeks of work)
Rain Paharia used Sonnet 4.1 to replicate map APIs and generate doc tests for her 'iddqd' library, reducing work that would have taken weeks into less than a day.
“I just told it to kind of replicate, you know, the same APIs across all of the other maps. … I just told the LLM to do that. … it took me, like, I want to say, like, less than a day to get, like, the whole thing ready.”
Rain Paharia · Hear it at 19:45
Engineering & simulationOpus 4.5found 3 real bugs
David Crespo used Opus 4.5 to analyze crash dumps and source code in the Ghostty terminal project, identifying three real bugs and writing bug reports without prior Zig expertise.
“I just have Opus essentially figure out. … it basically was able to, like, look at the code, statically analyze it, and find the source of these of I found three different bugs this way.”
David Crespo · Hear it at 31:04
Engineering & simulationOpus 4.52-3 hours (vs likely dropping the bug)
Rain Paharia used Opus 4.5 to investigate a SIGTTOU bug in nextest by analyzing source code from other projects, leading to a complete fix in about three hours.
“I I kind of gave this to OPUS four dot five. … it spent like ten minutes or so, and it actually wrote up a nice summary … it took me like maybe two or three hours to do.”
Rain Paharia · Hear it at 52:25
Engineering & simulationOpenAI models (VS Code)
Adam Leventhal used OpenAI models in VS Code to autocomplete an OpenAPI diff library and a CI tool, allowing him to build and demo the software in a short time.
“I started using some of the open API open API excuse me, OpenAI models in Versus Code … literally the rest of the program, I just had tab completed.”
Adam Leventhal · Hear it at 57:27
Engineering & simulationClaude Opus 4.5
Rain Paharia used Claude Opus 4.5 to identify verification properties for the Kani model checker in a Rust project, accelerating her learning and application of formal verification.
“I just asked CloudOpus four dot five to, hey. Like, you know, come up with a few properties that we can can verify that way. And it just did that.”
Rain Paharia · Hear it at 1:22:08
companies scaling up
Other shows on the same ground, closest first.
Jeff Gable, Luca, and Ryan Torvik discuss integrating Large Language Models (LLMs) into embedded systems engineering workflows, emphasizing that AI must augment rather than replace rigorous engineering discipline. They identify 'vibe coding' as a critical anti-pattern where developers lose control of architectural decisions by letting AI generate too much code at once, leading to unmanageable outputs and hallucinations. The hosts advocate for strict context management, test-driven development (TDD), and small, incremental commits to maintain deterministic quality, particularly in safety-critical domains like medical devices and automotive systems. They conclude that while LLMs accelerate prototyping and boilerplate generation, human code review and ownership remain non-negotiable for reliable production software.
Luca, an embedded software expert, discusses his experience at the Embedded Online Conference and a German conference, highlighting an LLM-only hackathon where participants were forbidden from writing code manually. He explains that effective AI-assisted development requires rigorous context management, treating prompts like version-controlled code, and managing external context files to prevent hallucinations. The conversation covers the economic benefits of AI (approx. $0.01 per line of code), the tightening of feedback loops, and the evolving role of developers from coders to AI supervisors and technical product owners.
François Baldassari, CEO of Memfault, and Thomas Sarlandie, Field CTO of Memfault, host a panel with Jonathan Beri, CEO of Golioth, and Dan Mangum, CTO of Golioth, to discuss the intersection of AI, open source, and embedded development. The group analyzes the maturity of edge AI for audio and video processing, the challenges of integrating LLMs into hardware due to rapid model iteration, and the current state of using AI for firmware development. They conclude that while AI is transforming code generation and review, open source remains critical for training models and providing trusted, complex infrastructure like Zephyr and FreeRTOS.
Ari Mahpour, an Embedded Systems Engineer at Rivian with prior experience at NASA JPL and Virgin Hyperloop, discusses the integration of AI and DevOps into embedded hardware development. He demonstrates 'vibe coding' with an Arduino using voice commands and LLMs, explaining how AI agents can autonomously write, compile, and test firmware. Mahpour also covers the challenges of managing hardware revisions, the importance of Model Context Protocol (MCP) for tool guardrails, and the future of AI in automated PCB design and testing.
Host Jeff Gable interviews Luca Johnny about integrating AI tools into embedded software development workflows. They compare IDE-integrated tools like GitHub Copilot and Cursor with command-line agents like Aider, emphasizing that workflow integration matters more than the underlying LLM. The discussion covers specific use cases for testing, debugging, and learning new frameworks, while warning that AI cannot replace the architectural judgment of experienced engineers.
The summary, chapters and topics are generated from the episode’s audio by this directory, not written by Oxide and Friends. The episode itself belongs to its publisher.