Transcription: OpenAgents Episode 194 - The Trillion-Dollar Question
June 9, 2026 ยท View on GitHub
Source: https://x.com/OpenAgentsInc/status/1986655639431500034 Wiki source: https://raw.githubusercontent.com/wiki/OpenAgentsInc/openagents/Video-Series.md Media title: OpenAgents - Episode 194: The Trillion-Dollar Question What percentage of global ... Upload date: 20251107 Transcription model: gpt-4o-transcribe-diarize Generated at: 2026-06-01T03:35:23Z
Machine-generated transcript. Review speaker labels and wording before using this as quote-grade source material.
[00:00] Christopher David: Yes, really, you can now code from your phone with Codex or Claude Code with a single NPX command,
[00:08] Christopher David: a single button to scan the QR code, and my devices are now paired.
[00:15] Christopher David: You'll see the log stream in and I'll say hello Codex, who are you?
[00:22] Christopher David: And you'll see the log streaming in. This is talking to your local codecs install. And you say, you know what? Codecs isn't being too smart today. Let's pop over to Claude code, new chat, Claude code.
[00:35] Christopher David: Who are you?
[00:39] Christopher David: And of course, Claude will kind of think about it for a second and give you an essay about.
[00:45] Christopher David: What Claude thinks and feels. Okay,
[00:47] Christopher David: so this is a app live now in test flight.
[00:53] Christopher David: Got the history there.
[00:54] Christopher David: I'm excited for developers to experience what I experienced last week where I had probably my single greatest increase,
[01:03] Christopher David: daily increase in my own personal productivity.
[01:06] Christopher David: Just getting this working, walking around my house and talking to it.
[01:10] Christopher David: Figured out how to...
[01:13] Christopher David: Get this QR code thing working sustainably for other people where now you have two options.
[01:18] Christopher David: If you run it like this, just with the basics and nothing else, there's no account needed.
[01:24] Christopher David: It just connects,
[01:25] Christopher David: it exposes this over your local area network so your devices need to be on the same Wi-Fi network.
[01:31] Christopher David: No account needed for that. If I go up here and I flip on my TailScale VPN,
[01:36] Christopher David: this same script will auto detect that I'm running TailScale.
[01:41] Christopher David: and i'll say tail scale and now it's the exact same flow but now i can take my phone elsewhere in the world and as long as i am my desktop is running i'll be able to chat with it from the grocery store wherever i'm at i don't think any other app enables that at least not in like a nice pretty mobile app like this when i say pretty i mean this code base is like
[02:03] Christopher David: a week old so the ui is still very rough but now that we've got the hard parts done we're just going to rapidly add features another kind of quality of life stuff one thing i want to add is like much better voice transcription probably via grok groq
[02:19] Christopher David: And that's about it. We're open to pull requests.
[02:23] Christopher David: All this is 100% open source,
[02:25] Christopher David: by the way.
[02:26] Christopher David: Thank you, Kia,
[02:27] Christopher David: giving us some stuff there.
[02:29] Christopher David: Nits are welcome, too, if they got good PRs attached.
[02:31] Christopher David: All this is 100% open source.
[02:33] Christopher David: I would not run anything like this if it was not 100% open source.
[02:37] Christopher David: What else? Live now on TestFlight.
[02:39] Christopher David: You can join the Android wait list by tweeting at us when Android.
[02:44] Christopher David: We got docs.
[02:46] Christopher David: that are decent,
[02:47] Christopher David: mostly AI-generated, but anything that is not clear that is not documented there,
[02:51] Christopher David: please DM or tweet at us.
[02:55] Christopher David: What else do we have? The tricoder script is on NPX. You need to be on 0.2.4 at minimum,
[03:01] Christopher David: but generally if you run NPX tricoder at latest, you'll get the thing that's pretty much guaranteed to work.
[03:08] Christopher David: I wanted to take a look at one little fun thing here,
[03:11] Christopher David: which was...
[03:12] Christopher David: Yeah.
[03:14] Christopher David: Open source successor to Super Maven.
[03:18] Christopher David: This little like lightweight completion thing that worked in Neovim and stuff.
[03:23] Christopher David: Acquired by Cursor.
[03:24] Christopher David: So they said when the blog post came out, yeah,
[03:27] Christopher David: SuperMaven will remain maintained,
[03:29] Christopher David: but it is now no longer going to be maintained. It's going to be sunset.
[03:33] Christopher David: Ironclad Dev here says this is my 9-11 and I will never forgive Cursor for this.
[03:39] Christopher David: Cursor, huh?
[03:40] Christopher David: Hey, Cursor, what's Cursor's valuation?
[03:44] Christopher David: Oh, see you soon, guys.
[03:46] Christopher David: The end of cloud inference?
[03:49] Christopher David: Maybe.
[03:51] Christopher David: People just remember,
[03:53] Christopher David: just remember that when people around T-Pont,
[03:57] Christopher David: AI Twitter,
[03:58] Christopher David: days or weeks from now talk about how cool the foundation model's API is from Apple,
[04:03] Christopher David: that you heard it here first,
[04:05] Christopher David: okay?
[04:05] Christopher David: For some reason, all of you are sleeping on it,
[04:08] Christopher David: fine, good for me.
[04:11] Christopher David: But Apple has put out amazing new primitives for app devs, rivaling and in some cases exceeding the Versal AI SDK in my own experience and what I can tell of its potential.
[04:25] Christopher David: Massively, massively impressed by it.
[04:27] Christopher David: And it seems to be working fine.
[04:29] Christopher David: This is the first ever that I've seen agentic search.
[04:34] Christopher David: through a code base, grep, and all of that stuff using tool calls.
[04:38] Christopher David: You can run it on device.
[04:42] Christopher David: I have it hooked up because we're doing a desktop mobile sync.
[04:44] Christopher David: I just run it on the M2 chip and then stream it over WebSockets to my mobile device for controlling QuadCode. But we're going to be supplementing QuadCode and Codex with the on device.
[04:59] Christopher David: model.
[05:01] Christopher David: Which got me thinking,
[05:04] Christopher David: Hey chat GPT,
[05:07] Christopher David: what percentage of AI workloads in the next five years can or will run on Apple silicon give me a range of possibilities and explore the ramifications for open AI,
[05:18] Christopher David: Nvidia and other major players if certain percentages of AI workloads move from cloud to the edge.
[05:24] Christopher David: So I found this at chat GPT pro.
[05:27] Christopher David: which can search the internet for sources and stuff like that. I also fed in recent tweets and analysis,
[05:32] Christopher David: a little bit of this, a little bit of that,
[05:34] Christopher David: Chris Pake's essay.
[05:37] Christopher David: And I'm going to share this link in the tweet below this. You'll see it.
[05:41] Christopher David: Summaries, blah, blah, blah.
[05:42] Christopher David: I've got a one-liner down here. If the next five years play out as current vectors suggest.
[05:47] Christopher David: Apple Silicon could end up running 15 to 25% of the world's AI inference by 2030 with a credible 7 to 31% total range.
[05:56] Christopher David: It gave like a stretch amount earlier of 35% also.
[06:00] Christopher David: And that swing is large enough to reshape GPU demand,
[06:03] Christopher David: API business models,
[06:05] Christopher David: and who captures the economics of everyday AI.
[06:12] Christopher David: And then I have to do a little bit of math about like what is the actual market cap implications of all that.
[06:18] Christopher David: Is it correct to call this a trillion dollar question?
[06:21] Christopher David: Yeah, yeah, easily a trillion dollar swing.
[06:24] Christopher David: Oops.
[06:25] Christopher David: Thumbs up.
[06:28] Christopher David: And what does it say up here for being able to tell if it's on track to being on the high end of that 7 to 30 percent?
[06:40] Christopher David: What to watch for. What would have to be true for the high scenario?
[06:43] Christopher David: Small, capable models. It's already got one of these. The one on there is, you know, like equivalent of 4B.
[06:51] Christopher David: I'll skip two for a second.
[06:52] Christopher David: PCC scale out.
[06:53] Christopher David: The sort of like central cloud that things could elevate to if you want a more powerful model.
[06:58] Christopher David: Hardware cadence.
[06:59] Christopher David: Okay.
[07:00] Christopher David: Now I don't have much control over 1, 3, and 4 there.
[07:03] Christopher David: Number two, however.
[07:05] Christopher David: Developer adoption of Apple's foundation models plus MLX is broad.
[07:11] Christopher David: None of you people have been talking about this at all.
[07:14] Christopher David: You didn't know,
[07:15] Christopher David: I guess.
[07:16] Christopher David: You're waiting for me to make a video about it. Is that right?
[07:19] Christopher David: Okay.
[07:21] Christopher David: Well,
[07:21] Christopher David: we're going to continue doing what we're doing, which is putting out open source releases on our GitHub.
[07:28] Christopher David: I'm going to continue building in public.
[07:30] Christopher David: We're going to continue making videos,
[07:32] Christopher David: and I'm going to chronicle for you over the next few videos over hopefully the next week or so what percentage of our own workload as I'm focusing on coding
[07:47] Christopher David: agents,
[07:48] Christopher David: specifically letting you manage the top coding agents, currently cloud coding codecs from your phone.
[07:57] Christopher David: It's like what percentage of our own coding agent workload is in the cloud?
[08:02] Christopher David: Well right now it's 100% but with version 0.3 of our Tricoder app in a few days,
[08:08] Christopher David: that's going to drop from 100% to I don't know, 95%. And in the beginning we're just going to be using the local model for a little bit of orchestration, we're generating title summaries but like 100 is going to drop to 95 and then it might keep dropping. What other things can we strip out?
[08:27] Christopher David: go local can we bring that down to fifty percent
[08:33] Christopher David: twenty percent at what point does that start having broader ramifications so stay tuned if you're a developer who wants to leverage the best coding agents from your phone go sign up for our test flight this is going to be iOS only for probably the foreseeable future next month or so at least
[08:54] Christopher David: Go sign up.
[08:55] Christopher David: If you're an angel investor or an early stage VC fund and you want exposure to this thesis,
[09:00] Christopher David: my DMs are open.
[09:03] Christopher David: Message me sooner than later,
[09:04] Christopher David: I recommend.
[09:05] Christopher David: See ya!