The Pragmatic Engineer: Mitchell Hashimoto’s New Way of Writing Code

Host: Gergely Orosz
Guest: Mitchell Hashimoto (co-founder of HashiCorp, creator of Terraform, Vagrant, and Ghostty)
Date: February 25, 2026

Episode Overview

In this deeply insightful episode, Mitchell Hashimoto shares his journey from self-taught teenage coder to co-founder of HashiCorp, offering rare, candid reflections on the evolution of open source, enterprise software, and the integration of AI into real-world workflows. The conversation spans formative experiences, lessons from building developer tools at scale, honest views on cloud providers, dealing with open source challenges in the AI era, and Mitchell’s philosophy and hands-on tactics for leveraging AI agents effectively.

Key Sections & Discussion Highlights

1. Early Days & Entry Into Coding

Timestamps: 01:30–07:54

• Mitchell started coding at age 12–13, motivated by video games, but quickly shifted to web, learning from whatever was freely available online and open source code.
• Self-education was enabled by online resources:

  “The only way I could learn was through whatever code was published online. And so that's how I got acquainted with open source. I didn't know that's what it was called then...” — Mitchell, 01:30

• Early project: reading the PHP manual daily on walks to school until variables “clicked.”
• Early freelance gigs masquerading as an 18-year-old on online marketplaces, driven by curiosity and necessity.

Notable Quote:

“I never became a video game programmer. I really quickly just became a web programmer, php, Perl, that sort of stuff.”
— Mitchell Hashimoto, 01:30

2. Loneliness & Finding Community Online

Timestamps: 04:12–05:50

• Coding was a solitary activity among school friends, but online chats and forums became a crucial support system.
• The breakthrough job came from a blog, leading to a chance Ruby on Rails consultancy position at 18.

3. Pathways Into Infrastructure & HashiCorp’s Origin

Timestamps: 06:29–12:18

• Mentorship at work: boss challenged him with no-mouse rule and progressively introduced devops tools, sparking long-term interest in infrastructure.
• Failed university project (Seattle Project) turned into a notebook of unsolved infrastructure challenges—a precursor to HashiCorp products.
• The notebook was key:

  “A subset of that was ultimately what Hashicorp would end up building. And I shared this with my undergraduate like boss who is Arman, who was my co founder.”
  — Mitchell, 10:40

• Quick decision to co-found a startup with Armon:

  “He emailed me that at like 11:30 near college. I emailed him back in two minutes and said, sure.” — Mitchell, 11:30

4. Vagrant, HashiCorp, and Betting Early on Multi-Cloud

Timestamps: 12:14–17:34

• Vagrant was born from pain-points in consultancy: repeatable dev environments.
• Chose VirtualBox for Vagrant due to being free—highlighting the role of constraints in engineering.
• Early experience with cloud (AWS) and the conviction that cloud (and multi-cloud) would be bigger than most assumed:

  “Anything that's economically huge, other people want a piece of that pie... you're going to have these others pop up and Microsoft is not going to sleep on it and Google's not going to sleep on it...” — Mitchell, 16:54

5. Building and Commercializing the HashiCorp Stack

Timestamps: 24:43–32:23

• Initial products: Vagrant, Packer (image-building), Consul (service discovery), Terraform (infrastructure as code), Vault (secrets management), Nomad (scheduling).
• Early open source focus—business model came later.
• Lessons from failed commercialization (“Atlas”): companies want per-product enterprise features, budgets are fragmented, and clarity of direction is highly valued:

  “We convened an all hands meeting... and said, ‘Okay, we're switching directions. We are now Enterprise as our customer...’ Nobody quit. The vibes in Slack were amazing. Super positive.”
  — Mitchell, 35:52

6. Insights on Cloud Providers: AWS, Azure, GCP

Timestamps: 58:19–65:13

• AWS: “Really arrogant... subtle vibe of ‘we will just spin up a product and kill your company.’”
• Microsoft: “Super competent professionals and team players... first to support Terraform.”
• Google Cloud: “Best technology, most incredible architectural thinking... but none of them cared or thought about the business at all.”

Notable Quote:

“By arrogant, I mean it always felt like they were doing us a favor at every turn in terms of partnerships... also this subtle vibe... we will just spin up a product and kill your company.”
— Mitchell, 59:39

7. The Changing Nature of Open Source in the Age of AI

Timestamps: 65:13–89:33

• AI contributions have massively increased noise and low-quality PRs in open source.
• Mitchell now requires strict disclosure of AI usage and has moved to a “vouching system” for PR submissions in Ghostty:

  “The disclosure worked decently. The issue was that the quantity of low quality AI PRs that we were getting reached a point where it was too high.” — Mitchell, 81:59

• Praises forking:

  “There should be a lot more forks. Like a lot more forks...the core privilege you get with open source, like osi, open source is forking and you should take that.”

• Predicts deeper changes in how open source communities must operate under relentless agentic AI contributions.

8. Harnessing AI Agents in Personal Workflow

Timestamps: 76:38–107:14

• AI has ironically boosted terminal usage via explosion of CLI/agentic tools.
• Mitchell always aims to have an AI agent doing something in the background:

  “What I try to do is I try, I endeavor to always have an agent doing something at all times. Maybe not when I sleep...but while I'm working, I basically say I want an agent. If I'm coding, I want an agent planning. If I, if, if they're coding, I want to be reviewing...”
  — Mitchell, 79:20

• Strictly manages interruptions: agents never interrupt, only Mitchell interrupts them, maintaining focus and deep work.
• Advocates reviewing all agent work on important projects; low-risk, time-limited tasks can be shipped with lower scrutiny.
• Stresses the need for engineers to develop proficiency in AI tools:

  “I would definitely require competency with AI tools. You don't need to use them for everything... It's like any other tool where sometimes it's useful and sometimes not...” — Mitchell, 102:59

9. Forecast: Codebases, Git, and Engineering Practices Will Evolve

Timestamps: 91:12–98:03

• AI drastically increases code churn, challenging git-based monorepos and merge queues.
• Expects major changes in version control, testing, CI/CD due to agentic code generators.
• Observability and sandboxing are further pressured by scale and the nature of AI-driven development.

10. Building, Hiring, and Personal Reflections

Timestamps: 99:06–116:36

• Best engineers often have "boring backgrounds"—private, not open source stars, but deeply competent, focused, and context-switching the least.
• Social media is “negative sum” for deep, focused engineering.
• For potential founders:

  “Imagine 10 years... is this really something you wanna work on for 10 years?”

• Personal recharging comes from introverted solitude, reading fiction, and family—not TV or public events.

Most Memorable Quotes

• On AI agents in his workflow:

  “I endeavor to always have an agent doing something at all times... I want an agent planning. If they're coding, I want to be reviewing... There should always be an agent doing something.” (79:20)

• On open source AI contributions:

  “AI makes it trivial to create plausible looking but incorrect and low quality contributions. And that's the fundamental issue.” (88:02)

• On the craft of software:

  “A lot of Ghosty is just the love of the game... Even though for end users it doesn't make a difference.” (76:09)

Timestamps for Key Segments

• Mitchell’s Coding Origin Story: 01:30–07:54
• Founding HashiCorp: 10:40–12:14
• HashiCorp Product Evolution: 24:43–31:18
• Enterprise Model Pivot (Atlas Failure): 34:47–40:53
• Experience with AWS, Azure, GCP: 58:19–65:13
• Open Source & AI PR Challenges: 81:12–88:02
• Managing AI Agents: 76:38–80:32
• Advice for Engineers & Founders: 99:06–111:28

Actionable Takeaways

• Engineers:

  • Get comfortable with AI tools as a core part of your workflow, not a replacement but a complement.
  • Structured, focused deep work still matters—don’t let agents fragment your attention.
  • Be ruthless about where code comes from and which contributions you invest time reviewing.

• Leaders/Founders:

  • When engineering for scale, embrace rapid iteration but always listen to customer signals.
  • Clarity, conviction, and willingness to dramatically change direction are essential.
  • Be prepared for fundamentally new bottlenecks as AI agents increase code output.

• Open Source Maintainers:

  • Consider innovations like vouching systems to handle AI-driven contribution floods.
  • Expect contributor expectations and “entitlement” to shift alongside more powerful agentic tooling.

Mitchell’s philosophy is pragmatic, hands-on, and relentless in its focus both on the craft of building and the necessity of adapting to dramatic technological shifts. This candid conversation is a must-listen for anyone who wants practical insight into the changing world of software engineering, from one of its most thoughtful builders.