Does AI Make Open Core Models More Difficult? with Sam Alba

Does AI Make Open Core Models More Difficult? with Sam Alba

Today on The Business of Open Source I spoke with Sam Alba [https://www.linkedin.com/in/samalba/], co-founder of Mendral [https://www.mendral.com], formerly of Dagger, and the first hire at Docker. We talked a lot about about how AI is changing the developer tools marketplace. If AI is changing the way software is developed, what does that mean for the companies that previously had made tools for developers?  We talked a lot about hype; what is hype and what is not hype? How many people are actually using AI and getting value from it.  It’s hard to differentiate between the hype and real value, Alba says, but also that while it does seem like AI is a bubble, that doesn’t mean there’s nothing real there. It’s just hard to see where the difference is now, but after the bubble bursts there will still be plenty of changes that stick around.  When it comes to building an open source company — which Mendral is not, by the way, Alba talks about how you have to get product-market fit twice. But for AI is absolutely changing things, and there have been companies that are abandoning open source because the game is changing so much and it is harder to make it work.  Listen to the entire podcast! And if you want someone to work with on establishing your message and then sharing that message in as many places as possible, get in touch.

Closed to Open Source Series: Doing Open Source "Right" with Ethan Arrowood

One of the myths that swirl around the open source company ecosystem is that all open source companies started as an open source project that then suddenly got traction, and followed by the founders frantically throwing spaghetti at the wall trying to figure out a business model.  In fact, this is not the story of how all open source companies came to exist.  It is not at all uncommon for a company to start the open source project with the explicit goal of building a company around it, and it also isn’t uncommon for a company that started as a closed source company to open source a portion of their codebase, for a variety of reasons.  This episode is about that second scenario. It’s the first in a series about how Harper went from a closed-source company to an open source company, including both the why and the how of the transition.  Today I talked with Ethan Arrowood [https://www.linkedin.com/in/ethan-arrowood/], head of open source engineering at Harper. [https://www.harper.fast]We’ll get his perspective on the transition from closed to open source today, and another perspective next time.  We talked about things like why Harper wasn’t open source from the very beginning, how Harper doesn’t fit the stereotype of an open source company in that they started with enterprise adoption, not individual developers using the platform, and the decision to go open source went hand in hand with the desire to focus on community building. We also had a conversation about the difference between being open source in name and really following the spirit of open source, and why there are business benefits to doing open source ‘right.’

The Value of Code is Dropping to Zero with Yann Lechelle

In this episode of The Business of Open Source, I talked with Yann Lechelle [https://www.linkedin.com/in/ylechelle/], Executive President of :Probabl [https://probabl.ai]. Perhaps the most interesting part of the conversation was about whether or not there is a risk from generative AI being trained on open source code, then generating new open source code and then re-training itself on that code. Yann says that there is already enough code in the world to give AI enough source material to work with, at least when it comes to ‘traditional’ code. So while perhaps there are issues with types of code that really require creativity, in most contexts we don’t need any more human-written code.  Does this mean the value of code is dropping to zero? That’s what Yann things… but for open source companies with permissive licenses, the value of code was already zero. This means that what really matters is community and brand.  What do you think? Do you think generative AI makes the value of code close to zero, and puts all software companies in the same boat as open source projects with permissive licenses? If you have an interesting take, let me know. And if you want someone to help you focus your positioning and streamline your product and marketing strategy, reach out. [https://www.emilyomier.com]

Why the World Needs the Agentic AI Foundation with Manik Surtani

In this episode of The Business of Open Source, I spoke with Manik Surtani [https://www.linkedin.com/in/maniksurtani/], one of the co-founders of the Agentic AI Foundation [https://aaif.io] and the CTO at the foundation. This is part of the series I’m working on about open source and AI, which started last week with a conversation with Glauber Costa about how AI killed a bug bounty program.  [https://thebusinessofopensource.transistor.fm/episodes/the-ai-induced-death-of-a-bug-bounty-program-with-glauber-costa] Manik talks about how the foundation came into existence, why it’s important to have a foundation that’s specific to agentic AI and what it means, in terms of everyday activities, to be the CTO of an open source foundation.  Given how fast everything is moving in the AI space, and specifically around what open source actually means, how we define what is and is not open source, where we can get data and who is able to have data that is open enough to be considered open source.  And if we want to mitigate the environmental impact of AI, is the solution really to insist on fewer cat videos?  However, if you like this show and want more content about the intersection of open source, AI and bottom lines, you should consider sponsoring! Reach out if you’re interested.

The AI-Induced Death of A Bug Bounty Program with Glauber Costa

The Business of Open Source is back! I’m starting a series about AI and open source this week.  I reached out to Glauber Costa [https://www.linkedin.com/in/glommer/], founder of Turso [https://turso.tech], after reading a post of his on LinkedIn [https://www.linkedin.com/feed/update/urn:li:activity:7460313925948162048/] about how bot-written PRs for their bug bounty program forced them to discontinue the program completely. In this episode, he talked about the bug bounty program — how it started, who contributed to it initially, why he considered it a huge success. And then he talks about what started happening when bots entered the picture.  He also talked about the difference between an open source project that accepts contributions and one that doesn’t, about the difference between an open source project and software that’s in the public domain, and how people in open source used to be seen as weirdos who hate money.  Glauber isn’t an AI hater — he talks about how they use AI at Turso, and how he has no problem with AI-assisted pull requests. The issue is when the result isn’t high-quality. There's also a difference between AI-assisted and 100% bot written. Then it creates essentially a denial of service attack on the community, because the maintainers end up having to spend so much time responding to bot-created PRs.  What’s your experience with AI and Open Source? Who else should I talk to? Let me know.  [http://www.emilyomier.com] Do you like The Business of Open Source? Help it to continue to exist by sponsoring the podcast. Does your company have a positioning problem? Work with me [https://www.emilyomier.com/work-with-me] to better position your product and see your growth take off.