The Future of Apps in an AI-Coded World | AppStories — Transcript

Full transcript

1. 0:00Hello and welcome to another episode of
2. 0:03AppStories.
3. 0:04I'm John Voorhees, and I have Federico to
4. 0:07teach you with me.
5. 0:08Hey, Federico.
6. 0:08Hello, John, how are you?
7. 0:09I'm doing great.
8. 0:10I'm doing great.
9. 0:11We got a good topic today.
10. 0:12We've got a lot of good topics, but
11. 0:14before we dig into the meat of the
12. 0:16episode, I wanted to do a very brief
13. 0:18little bit of follow-up with you, because
14. 0:19we talked all about Clawdbot on Mac
15. 0:22stories unwind of all places a couple of
16. 0:24weeks ago.
17. 0:24And then I know you also talked with
18. 0:27Steven and Mike about it on connected a
19. 0:29couple of times, and I thought I would
20. 0:32catch everybody up here.
21. 0:33And so what's been going on because it's
22. 0:35been evolving so rapidly.
23. 0:37And an app story seems like the best
24. 0:39place for it.
25. 0:39We don't have to repeat everything we
26. 0:41talked about in unwind or you talked
27. 0:43about unconnected.
28. 0:44But where is everything?
29. 0:46And what should people be thinking about
30. 0:49this project right now, I guess, is what
31. 0:51I would say.
32. 0:51So it's been renamed twice?
33. 0:54So obviously following the original name
34. 0:56of Claude Bot.
35. 0:57Anthropic got in touch when the project
36. 1:00started to get really popular and they
37. 1:04asked, since the name was very similar
38. 1:06sounding to Claude, they seem to have
39. 1:09asked, I don't know if politely or not
40. 1:12for a name change.
41. 1:14And so in a, according to creator Peter
42. 1:16Steinberger, um, at 5 AM in a sleep
43. 1:20deprived, sort of brainstorming session
44. 1:22on discord of all places.
45. 1:24The community voted for this name
46. 1:26Maltbot.
47. 1:28The idea that since the mascot was a crab
48. 1:32like a lobster, I think.
49. 1:34Like a lobster.
50. 1:35And lobsters, when they shed their skin
51. 1:38they molt, right?
52. 1:39And so they kind of went for this tongue
53. 1:43in-cheek name of malt bot, which was
54. 1:46horrible. to be fair.
55. 1:48But it was under pressure with a letter
56. 1:49from Anthropic.
57. 1:50And so I understand, you know, if you're
58. 1:52sleep deprived and you got a letter from
59. 1:54a huge corporation, with lawyers behind
60. 1:56it, and you're just a solo indie
61. 1:59developer running an open source
62. 2:00project.
63. 2:01I understand why you went for the rushed
64. 2:03decision.
65. 2:04But after a few days and lots of people
66. 2:06think, you know, boy, malt pot is really
67. 2:08a horrible name.
68. 2:09They landed in a not so sleep-deprived
69. 2:13state on the much better name as we're
70. 2:17recording this on Friday of OpenClaw.
71. 2:19So that's a much better name and it keeps
72. 2:22the original club thing with the lobster
73. 2:25and the sort of similar sounding to the
74. 2:28original club bot name.
75. 2:30But it's open source.
76. 2:31So OpenClaw.
77. 2:32I think it sounds much, much better.
78. 2:34It's a, it's still a funny name, but it's
79. 2:37also a normal name that is not malt bot.
80. 2:39All you have to do is go to the discord
81. 2:43for what was clawed bot is now OpenClaw
82. 2:46and you will realize... the chaos. 40,000
83. 2:50people.
84. 2:51It is, I've, I've, this, this is like
85. 2:54going to like a YouTube, a really big
86. 2:57YouTube live stream and looking at the
87. 2:59comments scroll up really fast.
88. 3:01I mean, it's almost to that level that
89. 3:03you cannot keep up with that discord.
90. 3:06There's so much going on there.
91. 3:06It is a huge project and it's really, I
92. 3:09mean, it's obviously a community now if
93. 3:11people contributing to the project.
94. 3:12But at the heart of it, it's just Peter.
95. 3:15Yeah.
96. 3:16Yeah.
97. 3:17And this project has really taken off in
98. 3:20a bunch of weird and unexpected
99. 3:22directions.
100. 3:22There's Cloudflare of all companies
101. 3:24Cloudflare.
102. 3:25That is now offering a managed, hosted
103. 3:28solution for OpenClaw.
104. 3:30Did you see John this other thing called
105. 3:33Malt book?
106. 3:34I did.
107. 3:35I was just reading about that on Simon
108. 3:36Wilson's block today.
109. 3:37A social network for the OpenClaw
110. 3:42agents.
111. 3:42So these agents have put together, like
112. 3:45the creator of Maltbook created this like
113. 3:48fake social network, like more like
114. 3:50Reddit, really.
115. 3:51It's like a fake Reddit for just AIs to
116. 3:54talk to each other.
117. 3:55And it's an insane place on the internet
118. 3:58where these agents are talking to each
119. 4:00other as if they were red users.
120. 4:02And they are creating their own
121. 4:04subcommunities.
122. 4:05They're creating their own like
123. 4:07narratives and stories and they're
124. 4:10sharing feelings.
125. 4:11They're not real feelings, mind you
126. 4:12they're fake AI agent feelings, and
127. 4:15they're talking to each other.
128. 4:16I mean, like, ah, yes, my human wanted me
129. 4:19to building something.
130. 4:21It's weird and creepy and fascinating and
131. 4:25also the data exfiltration risk of that
132. 4:28place.
133. 4:29Like, imagine hooking up your agent and
134. 4:31telling your agent, go have fun and hang
135. 4:33out on social network.
136. 4:35Go hang out on this random website that
137. 4:37anybody can post anything to, basically.
138. 4:39And imagine if like your agent starts, you
139. 4:42know, for the funzies of it all, starts
140. 4:45sharing your secrets, right, on this
141. 4:47social network, because that is still
142. 4:49your agent running your computer.
143. 4:51Or posting some JavaScript or something
144. 4:52that then does something to your
145. 4:54computer.
146. 4:54I mean, there's there's a lot of bad
147. 4:56things that could happen.
148. 4:57Simon Willison writes, given the inner end
149. 4:59risk of prompting injection against this
150. 5:02class of software, it's my current pick
151. 5:04for most likely to result in a challenger
152. 5:06disaster, but I'm going to put that aside
153. 5:09for the moment.
154. 5:09And those risks are real, even putting
155. 5:13aside something like this malt book thing
156. 5:15because there are, I know you've kind of
157. 5:18locked yours down a little bit more than
158. 5:19you had it originally locked down just
159. 5:21because your article has gotten so
160. 5:23popular and your name is all over all
161. 5:25over the internet with attached to Claude
162. 5:27Bot.
163. 5:28It's wild, man.
164. 5:30I really don't know.
165. 5:31But this project is obviously taken off
166. 5:33OpenClaw because it's, it's, it's, for
167. 5:36some people it's hitting all the right
168. 5:37notes in terms of what people want from
169. 5:40an AI powered assistant.
170. 5:42And at this point, given the popularity
171. 5:45of this, given how my article went really
172. 5:48popular.
173. 5:49Much more popular than I expected.
174. 5:51I really don't know where this is going
175. 5:53to land.
176. 5:54I really don't know.
177. 5:56There's probably a security disaster.
178. 5:58I think Simon Willis in Ira is right.
179. 6:00There's going to be some kind of disaster
180. 6:01at some point.
181. 6:03Be very mindful of what you hook up to
182. 6:06your open club agent.
183. 6:08Be very mindful of prompt injection and
184. 6:12data.
185. 6:14Make sure you know what are in this.
186. 6:15Make sure you know what's in the skills
187. 6:16both the scripts and the markdown.
188. 6:18Make sure you know what you're installing
189. 6:21from skills from other people.
190. 6:24Don't put it on your main computer if you
191. 6:26can...
192. 6:26I mean you should put it on a separate
193. 6:28computer if you can.
194. 6:28Federico did that with a mini.
195. 6:31Yeah, but I really don't know.
196. 6:34And I think at some point, you got to
197. 6:36wonder like, why is it so popular?
198. 6:38If you're OpenAI or if you're Anthropic
199. 6:41you must be wondering why is this thing
200. 6:43so popular?
201. 6:43Like, and I think, and I think there's
202. 6:45something there, if you're one of those
203. 6:47companies, you got to think about like
204. 6:49okay, maybe, maybe this proves that we
205. 6:51haven't been doing enough with our, with
206. 6:54our LLMs.
207. 6:55Because I think I think beyond the funny
208. 6:58story and now the fake social network
209. 7:02behind the memes and everything.
210. 7:03There is an underlying truth, that is
211. 7:06people are using this because it's
212. 7:07useful.
213. 7:08And so if you're one of those companies
214. 7:10you got to be thinking about that.
215. 7:11Like that core truth that it's useful.
216. 7:13And how can they make ChatGPT or cloud
217. 7:16more useful like that while being more
218. 7:19secure and safer and easier, obviously, to
219. 7:21get started?
220. 7:22But there's something there.
221. 7:23And I think that's the space that I want
222. 7:25to observe.
223. 7:25Yeah, same here.
224. 7:27I mean, do you think the difference
225. 7:29between this and, say, just having the
226. 7:32clawed app on your on your iPhone, is the
227. 7:36Continuity of it, the fact that it's
228. 7:38running on it?
229. 7:39It's got to be, I mean, it's, it's the
230. 7:41fact that it's running on a server
231. 7:42locally has access to your file system
232. 7:44and has some Continuity from session to
233. 7:47session, right?
234. 7:48Isn't that kind of like at a basic level
235. 7:50why this is more popular than just a one
236. 7:52off chatbot?
237. 7:53The reason why it's popular is that I
238. 7:56think part of it is the self-improvement
239. 7:59of it all.
240. 8:00Yeah, that's true.
241. 8:01The future doesn't, a feature does not
242. 8:03exist.
243. 8:03I'm just going to make myself a feature
244. 8:05for that.
245. 8:05And I also think, you know, I ran
246. 8:08something last night.
247. 8:09I, I, um, I, I wanted to see, now, it
248. 8:13seems like if, if, you know, if you spend
249. 8:15some time following some people, there
250. 8:17are still only posting on X, formerly
251. 8:19Twitter, you must have seen that now
252. 8:21suddenly X articles are really popular
253. 8:25like people are blogging on X. Yes, like
254. 8:281000 words. essays.
255. 8:30Jesus, but sure.
256. 8:32Let's go with that.
257. 8:34And I wanted to see if there's, is there
258. 8:37a way for me to get these articles?
259. 8:39Some of them are like this hype articles
260. 8:42that have basically no vibe.
261. 8:44But you occasionally find like a good how
262. 8:46to...
263. 8:46Very occasional gem.
264. 8:47And I was like, okay, there's this
265. 8:49interesting article from somebody who
266. 8:50works on Anthropic.
267. 8:51Is there a way for me to extract this
268. 8:54article from X via some API or something
269. 8:57to put the text somewhere else?
270. 8:59And I asked to both open Claude and
271. 9:02Claude, the regular Claude app in a
272. 9:04regular chat.
273. 9:05It immediately ran out of context.
274. 9:07And it started looping on itself. like
275. 9:11how do I do this?
276. 9:11And then like compaction fails.
277. 9:13You can no longer have chatting this
278. 9:16session.
279. 9:17I went to OpenClaw and because it's
280. 9:19based on the Claude Code, agantic harness.
281. 9:23It just did it.
282. 9:25It just found a way, eventually, you know
283. 9:28what it did.
284. 9:28What did it do?
285. 9:29It opened Google Chrome.
286. 9:30Of course it did.
287. 9:32Read the article.
288. 9:33But that's the thing.
289. 9:35That's the thing.
290. 9:35You are so limited in the regular chat
291. 9:38bots.
292. 9:38And I think that's what's resonating with
293. 9:40people.
294. 9:41OpenAI knows this.
295. 9:42They've been blogging about the fact that
296. 9:43they realize, oh, we made ChatGPT so
297. 9:45powerful and yet it's not that powerful.
298. 9:48Yeah, the overhang.
299. 9:52One of my favorite expressions of 40
300. 9:54years so far.
301. 9:55One of the 1st one of the 1st time I read
302. 9:57that, I was like, wait, what?
303. 9:58Is that English?
304. 9:59Anyway, anyway, OpenClaw.
305. 10:04Let's go OpenClaw now.
306. 10:06And it's funny because in our in our
307. 10:08analytics, I see so much traffic coming
308. 10:11from 3 separate domains now.
309. 10:12Clawdbot, Maltbot, an open clot, because
310. 10:16MacStories is linked on all the 3
311. 10:18domains.
312. 10:19Oh, that's funny.
313. 10:21Yeah, that's that's great.
314. 10:23That's great.
315. 10:23Okay, well, let's transition over. to the
316. 10:26main topic today.
317. 10:27And this is kind of what I would call the
318. 10:29wrap-up episode of our projects over the
319. 10:34holidays and then like tips and tricks
320. 10:37about how we're using things like Codex
321. 10:39and Clawdbot, um, not, well, Clawdbot and
322. 10:42just Claude Code.
323. 10:43And today, I want to talk about something
324. 10:45that I know that's been on your mind and
325. 10:47my mind a lot for many weeks now, which
326. 10:50is what does this all mean in terms of
327. 10:53the kind of things that we cover?
328. 10:56What does this mean in terms of apps?
329. 10:58Both the App Store itself?
330. 11:00developers that we know?
331. 11:01Where is this all heading?
332. 11:03And I think that that is something that
333. 11:06we probably aren't going to be able to
334. 11:07answer completely today.
335. 11:09But I think we both have a lot of
336. 11:11thoughts and feelings about it.
337. 11:12So, I don't know.
338. 11:14I'll let you kick it off, Federico.
339. 11:16What do you think?
340. 11:16Where do you want to start?
341. 11:17I feel like, I'm sorry.
342. 11:20I'm going to already interrupt you.
343. 11:21I think we need context.
344. 11:22You're just interrupting yourself, man.
345. 11:24I haven't said anything yet.
346. 11:27Well, I wanted to, one thing I wanted to
347. 11:29kind of lay out at the beginning of this
348. 11:32is that I get the sense sometimes that we
349. 11:38hear from readers who are like, well, I
350. 11:40won't, I'm not going to get any apps that
351. 11:43are using AI in the development process.
352. 11:46And I'm just kind of here to tell you
353. 11:49that it's already happening.
354. 11:51It's already happening probably with all
355. 11:52the apps that you're that you're using on
356. 11:54the App Store.
357. 11:55I think that this is a tool that what I'm
358. 11:57saying is this is a tool that people are
359. 12:00using and they're using it effectively.
360. 12:03If it's put it in the hands of a
361. 12:05developer who knows what they're doing, it
362. 12:07makes them more efficient.
363. 12:08And I think you don't hear about it a lot
364. 12:11necessarily as a user.
365. 12:12We hear about it from people we know, but
366. 12:15because AI can be so controversial in
367. 12:17some quarters.
368. 12:18I think people don't talk a lot about
369. 12:21what the tools they're using to build
370. 12:24their apps.
371. 12:24Yeah, and but it's happening.
372. 12:27I mean, even if you think you're taking
373. 12:29an anti-AI stands, which is perfectly
374. 12:33reasonable, but you should not live under
375. 12:37the impression that you're using
376. 12:39completely AI free software.
377. 12:41Unless you're literally spending time at
378. 12:43each developer's house and sort of
379. 12:45hanging over their shoulder and watching
380. 12:46them type code, or unless you are a
381. 12:49programmer yourself and you just
382. 12:50bootstrap your entire operating system.
383. 12:53Because, I mean, even the OS, Apple is
384. 12:56using...
385. 12:57People internally at Apple are using it.
386. 12:58Yes, absolutely.
387. 12:59So you cannot explain...
388. 13:00You cannot escape the cycle.
389. 13:02But that being said, um, I'll start with
390. 13:05the simple truth.
391. 13:06I think, um, I think, um, we are a the
392. 13:11beginning of this new era, and very
393. 13:13simply, there still hasn't been a quote
394. 13:17unquote App Store moment for vibe coded
395. 13:22software, whatever you want to call it.
396. 13:23I guess what I'm trying to say is that
397. 13:25doing this, writing your own personal
398. 13:28software, using AI to create something
399. 13:31that does not exist on the App Store, it's
400. 13:33still too hard to do, still too difficult
401. 13:36for most people.
402. 13:38I wouldn't go to my friends in real life
403. 13:39and telling, I wouldn't go there and
404. 13:41evangelize Claude Code to them.
405. 13:43This is still very much a hobbyist or
406. 13:46nerd or whatever you want to call it kind
407. 13:48of thing.
408. 13:49It's not for the masses, but it's going
409. 13:52to be.
410. 13:52That's the thing.
411. 13:53So it's better to start thinking about
412. 13:55the problems now, before the problems
413. 13:57arrive, all of a sudden, and we're left
414. 14:00we just staring at the ocean of vibe code
415. 14:03software.
416. 14:03I think it's better to think about the
417. 14:05ramifications now because it is going to
418. 14:07happen.
419. 14:08It's difficult now.
420. 14:09It's not for everybody, but it's getting
421. 14:11easier by the day, not even by the week
422. 14:14not even by the month.
423. 14:15It's getting easier by the day.
424. 14:16And so, um, that's the 1st thing I will
425. 14:19say.
426. 14:19Yeah, and I want to say that I think the
427. 14:211st place we're going to see this
428. 14:22Federico, is in the kind of single
429. 14:24purpose utilities.
430. 14:26Yes.
431. 14:27Because that's like you and I have been
432. 14:29essentially building our own single
433. 14:31purpose utilities for things that we do
434. 14:33for MacStories.
435. 14:34And that's essentially what we're doing.
436. 14:36We're taking, we're building what are
437. 14:38basically scripts, and we're wrapping
438. 14:41them in either like a web app, or in my
439. 14:43case, I've done a couple of native apps
440. 14:45as well, and doing it that way.
441. 14:47And that, to me, having gone through the
442. 14:49experience of building a couple of
443. 14:51different things, both on the for web
444. 14:52apps and for iOS and macOS, it's clear to
445. 14:56me that we're going to see an absolute
446. 14:59flood of single-purpose utilities.
447. 15:01So I think in the short term, we're going
448. 15:03to see a lot more apps on the App Store.
449. 15:05But I also think that additional supply
450. 15:08and the fact that people are building
451. 15:10their own tools, you know, the supply is
452. 15:13going to increase, the demand is going to
453. 15:14go down, and we're going to see a real
454. 15:17inversion there pretty quickly, I think.
455. 15:21Yes.
456. 15:22Yeah, yeah. that's I completely agree.
457. 15:24I think I saw this expression being used
458. 15:27online, abundance of software.
459. 15:30We live in a software abundance era.
460. 15:33If you are, all of these, all digital
461. 15:37revolutions, they start with computer
462. 15:40nerds.
463. 15:41And then they get simplified and
464. 15:43productized and consumerized and they
465. 15:47become easier for everybody to use.
466. 15:48But they always start with nerds.
467. 15:50They always start with people who listen
468. 15:52to AppStories with people who hang out
469. 15:53on mastodon, those kinds of people.
470. 15:55They start with programmers.
471. 15:56And then they get easier and easier and
472. 15:58easier and they become the equivalent of
473. 16:00a product you can buy off the shelf.
474. 16:02And I think right now what we're seeing
475. 16:04in our community, with people like us, is
476. 16:08that there's been a decline for, I used
477. 16:14to, I'm going to give you a very simple
478. 16:15example.
479. 16:16I used to regularly post, used to be
480. 16:19Twitter, then Mastodon, and Blue Sky, used
481. 16:22to regularly post, like, hey, does this
482. 16:23app exist?
483. 16:24Like or somebody should make this very
484. 16:27specific thing.
485. 16:27Now, I can make my own very specific
486. 16:30thing.
487. 16:31Yep.
488. 16:31And I can tell you this has been
489. 16:34happening to me.
490. 16:34Like I, uh, a few days ago.
491. 16:37A few weeks ago, I wanted to have a
492. 16:40simple utility just to append text or
493. 16:42audio or images to my daily noting
494. 16:45notion.
495. 16:45Like, what are the odds now that
496. 16:47something like that exists on the App
497. 16:49Store?
498. 16:49I would say pretty likely, right?
499. 16:51Because notion is popular, and there's a
500. 16:52pretty good chance that I could find
501. 16:54something to upend some text or some or
502. 16:57an image or some audio to a noting
503. 16:59notion.
504. 17:00But then I started thinking, well, but
505. 17:02would those daily notes be named exactly
506. 17:04like I want them to be named?
507. 17:06Exactly.
508. 17:06You start thinking like, what about all
509. 17:09my own little preoccupations that I
510. 17:11have?
511. 17:11All my little preferences that I have?
512. 17:13And so I ended up making one with clock
513. 17:15code and then Codex, as we talked about.
514. 17:18But that's what's changing right now.
515. 17:21That's the big difference, Federico, is
516. 17:24that you, if you had found that app, you
517. 17:27would be molding the way you work too, the
518. 17:29way the app works, whereas now you don't
519. 17:31have to do that.
520. 17:32You've already got your setup with your
521. 17:33daily note.
522. 17:34Now you can build something that works
523. 17:36the way you want it to with your existing
524. 17:38system.
525. 17:39You don't have to change anything.
526. 17:40Yes, yes.
527. 17:41And, um, so I think it's going to happen
528. 17:441st 2 utilities, to self-contained very
529. 17:46clearly defined, uh, little pieces of
530. 17:49software that are supposed to do, you
531. 17:51know, the age old expression one thing
532. 17:53well.
533. 17:54Now you're going to make it yourself.
534. 17:56Now, if you're a developer, what does
535. 18:00that mean?
536. 18:00I think, what we're gonna see next.
537. 18:03So I think I completely agree.
538. 18:05We're going to see an abundance of this
539. 18:06software, and there's going to be
540. 18:08problems with that, right?
541. 18:09in terms of copyright, like who owns the
542. 18:12code?
543. 18:12Nothing and everybody.
544. 18:14It's very strange.
545. 18:16But also there's going to be, I think at
546. 18:19some point we will get the 1st stories of
547. 18:23this vibe coded app that I got from the
548. 18:26App Store, leaked all my data because the
549. 18:29security practices were not strong
550. 18:30enough.
551. 18:31So there will be at some point a movement
552. 18:34toward using real developer made apps
553. 18:39again.
554. 18:41And there's still maybe people who, there
555. 18:43are already people who say that.
556. 18:44But I think if you're a developer, what I
557. 18:48would consider is, um, A different
558. 18:53paradigm for writing software.
559. 18:56And by that, I mean, it used to be very
560. 18:59deterministic, right?
561. 19:00That your developer, you want to make an
562. 19:03app for, I don't know, cropping pictures
563. 19:05uh, and you release it on the App Store.
564. 19:08And you recently the App Store and you
565. 19:09got a bunch of reviews and some people
566. 19:11like it and other people say, well, I
567. 19:13like it.
568. 19:13I like that I can crop pictures, but I
569. 19:16wish I could also duplicate pictures.
570. 19:17And your app does not duplicate pictures
571. 19:19right?
572. 19:20And so you got to go back in and add a
573. 19:22feature to duplicate pictures.
574. 19:23I think going forward.
575. 19:25We will see the rise of a new type of app
576. 19:30that is, you know, talking about open
577. 19:32claw, the kind of a self-improving app
578. 19:35where you as the user can go into this
579. 19:40type of application, and you can just add
580. 19:42a feature that does not exist yet.
581. 19:45Now that requires a fundamental
582. 19:48rethinking of the sandboxing model and
583. 19:51the technology model of an app.
584. 19:52Like, how can you rewrite something that
585. 19:54is not shipped on the App Store, like the
586. 19:59original binary of the app, right?
587. 20:00You're basically modifying software.
588. 20:02And I don't know the answer to that.
589. 20:04And I think maybe it could be something
590. 20:07you know, behind the scenes could be
591. 20:09something like running a script or
592. 20:11something like Swift playgrounds where
593. 20:14you can install Swift playgrounds, but
594. 20:16then you can make your own playground on
595. 20:17device.
596. 20:18But I think, uh, the rise of this new
597. 20:20type of app, of, of, uh, an agantic type
598. 20:24of app where, where the users very simply
599. 20:27can modify the experience and make them
600. 20:30their own without having to go to clot
601. 20:32code without having to sign up for
602. 20:34developer account.
603. 20:35I think if I were developer, I would
604. 20:37start considering this, which is not too
605. 20:39dissimilar. from, you know, editorial or
606. 20:42Pythonista or what these apps used to do
607. 20:45back in the day or even obsidian for that
608. 20:47matter, where if a feature does not exist
609. 20:49you can make your own plug-in.
610. 20:50I think we're going to see the rise of
611. 20:52that. that's a good point.
612. 20:54That's a very good point.
613. 20:55There are a couple of things here that
614. 20:58I've been thinking about.
615. 20:59And one of them is in building my own Mac
616. 21:03app over the last couple of weeks.
617. 21:06What I found in using Claude Code was
618. 21:09that I got 80 +% of the way there very
619. 21:14very quickly, like, you know, in a few
620. 21:17hours worth of fiddling around.
621. 21:18And then the process has been very, very
622. 21:21slow as I try to refine it and make it
623. 21:25more, make it more appealing to a mass
624. 21:30audience. you know, someone beyond just
625. 21:32myself.
626. 21:33And I think that that's kind of an
627. 21:35interesting thread here that I've been
628. 21:37thinking about a lot because on the one
629. 21:39hand, you and I can build a Python script
630. 21:42or and, you know, tie it up to a web app
631. 21:44and it'll work just the way you want it
632. 21:46or just the way I want it.
633. 21:47But that only works for us.
634. 21:51And where I think that there's still
635. 21:54going to be opportunities for developers
636. 21:57is in the design and the productizing of
637. 22:01some of these, uh, some of these
638. 22:04utilities and other, you know, other
639. 22:05other types of software because, you know
640. 22:09we're talking about people like you and I
641. 22:10who are willing to do this kind of work
642. 22:12to get exactly what we want.
643. 22:14We got to remember that the App Store is
644. 22:17still a place where most people won't buy
645. 22:19anything.
646. 22:19And most people don't download apps
647. 22:22obsessively like we do.
648. 22:24And so there's still a market for people
649. 22:27who don't want to get their, you know, get
650. 22:29their hands dirty in the code in a way, I
651. 22:31think.
652. 22:31And I think that that's not going to
653. 22:32change just because it's possible to do
654. 22:35it more easily.
655. 22:35I think there's still going to be a
656. 22:37decent, you know, good market.
657. 22:39The thing that I think, though, is that
658. 22:42you know, like the tension there is that
659. 22:44the people who really want, and this is
660. 22:46why I've been thinking about utility apps
661. 22:47a lot.
662. 22:48The people who really want a good single
663. 22:49purpose utility are the people who are
664. 22:51most likely to build it themselves
665. 22:53because they're very picky and particular
666. 22:55about what they want and how they have
667. 22:56their workflows set up.
668. 22:58And that, I think, is where a lot of the
669. 22:59initial play in this market is going to
670. 23:01be is with that, that market that I think
671. 23:05a lot of people, a lot of developers are
672. 23:07looking at and selling to, of people who
673. 23:09actually spend money on apps and actually
674. 23:12care about apps, and the thing that's
675. 23:14going to be different.
676. 23:15For them, between something they can
677. 23:18build themselves and something somebody
678. 23:19else builds for them is someone who takes
679. 23:22the time to kind of make it really nice.
680. 23:25And that's, you know, what, and really
681. 23:27nice is both from a product standpoint, a
682. 23:31design standpoint, all those kind of
683. 23:32things.
684. 23:33Yeah, yeah.
685. 23:34I think, obviously, the other...
686. 23:40The other idea is that maybe not
687. 23:43necessarily everybody thinks about
688. 23:45software along those lines.
689. 23:47You sent me these article a few days
690. 23:49ago.
691. 23:49What's it called?
692. 23:51Claude Code psychosis or something like
693. 23:53that.
694. 23:53This idea that while people like us in
695. 23:57our community are obsessing over these
696. 23:59new capabilities of agents like cloud
697. 24:01code and codex to make software on demand
698. 24:04personal software for you in just a few
699. 24:06minutes, Most people most people in real
700. 24:10life, they don't obsess over software
701. 24:13like that.
702. 24:13See problems as software problems, I
703. 24:16think, is one of the points of the
704. 24:17article, right?
705. 24:18Yeah, that's what the article says.
706. 24:19They don't see problems as software
707. 24:21problems all the time.
708. 24:22Whereas you and me were doing something
709. 24:23and that's the automation mindset, right?
710. 24:26We do something because we are knowledge
711. 24:28workers, as they say.
712. 24:30And overworked people who want to find a
713. 24:33shortcut.
714. 24:33Yeah, and but we, our brains are sort of
715. 24:37wired to think like that where you see a
716. 24:40problem and you see that problem twice or
717. 24:423 times and our brains immediately go, oh
718. 24:45I can, I can fix that with software
719. 24:47right?
720. 24:48But most people don't think like that.
721. 24:50Most people in real life, you know, that
722. 24:51don't have jobs that involve staring at a
723. 24:54terminal or keeping up with AI news.
724. 24:56If they don't like their notes app, they
725. 24:59don't think like, oh, maybe I should make
726. 25:00my own notes app.
727. 25:01They're just going to find something else
728. 25:03or complain about it.
729. 25:04Or live with it because most people have
730. 25:06a surprisingly high threshold for crappy
731. 25:08software.
732. 25:09Um, and, um, so that's why I want to go
733. 25:13back to the idea of the App Store
734. 25:15moment.
735. 25:15Like, where, how, what's the shape of a
736. 25:19moment in time, maybe in the not so
737. 25:22distant future where this idea, the core
738. 25:26of this idea, which is software can be
739. 25:29personalized per user?
740. 25:32The moment that it becomes so simple and
741. 25:34so intuitive, then anybody can do it.
742. 25:37And I'm not sure what that looks like.
743. 25:40I think it looks a little bit like what
744. 25:42we talked about at the very top of the
745. 25:44show, which is OpenClaw, where you have
746. 25:46an agent that does it for you, where it's
747. 25:51and it, but obviously it has to be, you
748. 25:54know, light years easier than what open
749. 25:56claw is, because that's a very fitly
750. 25:58process to set up.
751. 25:59But I think this is kind of where the
752. 26:02notion of everything apps are coming from
753. 26:04from a lot of these AI companies.
754. 26:06They want their AI agent to build the
755. 26:10software for you, that you'll go into
756. 26:12OpenAI's app, whatever it happens to be
757. 26:16or Anthropics.
758. 26:18And you'll just tell your agent, you know
759. 26:20I need something that'll help me fill out
760. 26:22my taxes or I need something to write me
761. 26:25a resume and it'll come up with something
762. 26:28on the fly and give it to you.
763. 26:30And it may be doing it from kind of a
764. 26:32templating system or it may be, you know
765. 26:36and then using that to craft something
766. 26:38more particular to your needs as you
767. 26:40describe it.
768. 26:40But I think it's gonna be something like
769. 26:42that.
770. 26:42Yeah, I think you're right.
771. 26:45And if anything, I think Of all companies
772. 26:49I think Apple has a pretty good shot at
773. 26:51doing this, even though they're late.
774. 26:54Um, They have.
775. 26:57I think they have a chance.
776. 26:58So they were like to the chatbot, right
777. 27:00to the chatbot UI, to the ChatGPT or
778. 27:02Claude or Gemini experience.
779. 27:04What they are not late to is this.
780. 27:09Like the idea of, yes, an assistant, yes
781. 27:13it's based on an, yes, it can be a
782. 27:16chatbot, but applying that power to user
783. 27:21personalization.
784. 27:22And I think whether that takes the shape
785. 27:24of, as we talked about, a fundamentally
786. 27:28rethought shortcuts, or something else
787. 27:32where maybe you're just talking to Siri
788. 27:35and since it's now much more capable, you
789. 27:37can just say things like, hey, Siri, can
790. 27:39you make a note that is, you know, has a
791. 27:43purple background and an agreed of photos
792. 27:46of my dogs and send that as a calendar
793. 27:50invitation for my dog birthday party. .
794. 27:52Isn't that personalized software?
795. 27:56It is personalized software.
796. 27:57And I think Apple has a chance to not be
797. 27:59late to that because nobody has figured
798. 28:02it out because nobody has the platform
799. 28:03and the app ecosystem and the devices to
800. 28:06do that.
801. 28:07So that will that would be a fascinating
802. 28:11move, especially if you're Apple, and, you
803. 28:14know, they invented the mobile App Store
804. 28:17and they still have a shot to redo it all
805. 28:21again from that angle, and I think it
806. 28:25would be really fascinating.
807. 28:26And obviously it would also be scary
808. 28:29because Apple makes a lot of money from
809. 28:31the App Store from services and from
810. 28:33commissions.
811. 28:34What happens when, would there be, would
812. 28:39they be incentivized to cannibalize the
813. 28:43App Store?
814. 28:44To ride this new wave of software?
815. 28:47And our developer's going to be
816. 28:48incentivized.
817. 28:49That's, I mean, I think the part here too
818. 28:50that it's like, I think from a user's
819. 28:54standpoint, things like app intents make
820. 28:56a lot of sense, and that's a good
821. 28:58foundation, which to build, on which to
822. 29:00kind of build this kind of model.
823. 29:01However, I do feel like from a developer
824. 29:06standpoint, if you have your
825. 29:08photocropping app and now all of a sudden
826. 29:11users can crop photos without ever having
827. 29:13to open your app, you know, you've
828. 29:16commoditized that functionality and there
829. 29:19could be 5 different apps on my phone
830. 29:21that crop things.
831. 29:22And yes, if you're 1st to the 1st to the
832. 29:25party by being the 1st one to implement
833. 29:27app intents, maybe that makes you the
834. 29:30default, but that's not always going to
835. 29:31be the case for every app that everyone
836. 29:33makes.
837. 29:33And so I do wonder, you know, how that
838. 29:35plays out in terms of developers wanting
839. 29:39to opt into this system or not.
840. 29:41I mean, we've seen a lot of developers we
841. 29:43know, build out app intents.
842. 29:46And I feel like we're still at a point
843. 29:49with app intents, where, It's there's a
844. 29:54lot of open questions still.
845. 29:55And I think the questions that were open
846. 29:57in 2024 when a lot of this was, I mean
847. 30:00app intents go back farther than that, but
848. 30:02when smarter Siri was 1st discussed, the
849. 30:06questions then have multiplied because
850. 30:09the world has changed since 2024.
851. 30:12And now, you know, how does this work
852. 30:14with Gemini being integrated?
853. 30:16There's just a lot of questions that
854. 30:19aren't really answered in terms of what
855. 30:21it means for developers ultimately.
856. 30:24Yeah.
857. 30:24Yeah, I don't know.
858. 30:27And there's, there's the, there's
859. 30:30obviously the, the developer angle also
860. 30:34from, from a financial perspective, like
861. 30:37if, if all of this is going to happen.
862. 30:39How are developers gonna make money?
863. 30:42Is anybody, is everybody just going to
864. 30:44become an AI developer?
865. 30:45Is everybody just going to be essentially
866. 30:47writing APIs for an AI?
867. 30:49That's, I mean, that's what it, you know
868. 30:50what I mean?
869. 30:51That's what that's what this is in a
870. 30:53way.
871. 30:53I mean, that's what we're seeing both on
872. 30:55the Anthropic and OpenAI side, but also
873. 30:58app intents are arguably that.
874. 31:00We talked about them being like endpoints
875. 31:02for APIs.
876. 31:03And that's great.
877. 31:04You and I love endpoints for APIs.
878. 31:05We can do all kinds of stuff with that
879. 31:07but that doesn't necessarily make it good
880. 31:09for developers.
881. 31:10Yeah, and that also turns into a bigger
882. 31:13conversation about like how he's
883. 31:15programming, changing in general.
884. 31:16You know, if, if, um, I saw, I think I
885. 31:21mentioned this on the show, I, uh, before
886. 31:23I saw somebody arguing that people who
887. 31:26say that AI is a bubble and it's going to
888. 31:29go away.
889. 31:30Do those people think that LLMs will be
890. 31:34uninvented?
891. 31:35Like, I mean, at this point, the cat's
892. 31:37out of the bag, right?
893. 31:38Even if the AI bubble bursts and OpenAI
894. 31:41goes bankrupt, an Anthropic ceases to
895. 31:45exist.
896. 31:45I mean, very unlikely scenarios, but
897. 31:47let's assume that it will happen.
898. 31:49Do we think that LLMs will suddenly
899. 31:52become irrelevant?
900. 31:53The web didn't go away when, you know
901. 31:55when all those companies crashed.
902. 31:57It didn't.
903. 31:58It didn't.
904. 31:58And so there's a question of if LLMS is a
905. 32:01programming, is replacing code written by
906. 32:05humans, what's the role of a human in a
907. 32:09programming economy going forward?
908. 32:11Are all developers just going to become
909. 32:13agent managers?
910. 32:15Like, is that is that what's going to
911. 32:16happen?
912. 32:16Um, like big picture?
913. 32:18Because even if you think, and when I say
914. 32:21big picture or long term.
915. 32:23I mean, what's a what's a long term at
916. 32:25this point?
917. 32:25Because a year ago, Claude Code did not
918. 32:27exist.
919. 32:27Right.
920. 32:28And and so what does he look what does he
921. 32:31look like a year from now?
922. 32:32I have no idea, but I see, I see, you
923. 32:37know, I take a look at people like Peter
924. 32:38Steinberger, for example, and Peter runs
925. 32:41open source projects, right?
926. 32:42They're not commercial projects.
927. 32:44Right.
928. 32:44But he's a former developer who sold his
929. 32:47previous company for like a $100
930. 32:48million.
931. 32:49So he knows programming.
932. 32:51He used to be a regular programmer by
933. 32:52trade 10 years ago.
934. 32:54And now it's basically a manager of
935. 32:57agents and sub agents and ships a ton of
936. 32:59open source projects.
937. 33:00And I would say at this point, it's fair
938. 33:03to imagine that maybe he mentioned that
939. 33:06the club bought website was visited by
940. 33:092000000 people.
941. 33:09So let's say that, well, a 1000000 people
942. 33:11installed Clawdbot or OpenClaw.
943. 33:15And he's just a guy managing multiple
944. 33:19terminal sessions with a bunch of agents
945. 33:21and subagents.
946. 33:22Is he a programmer?
947. 33:24I mean, obviously he's a programmer, but
948. 33:26he's a new type of programmer, and if we
949. 33:28extrapolate from that, and sort of launch
950. 33:32ourselves one year or 2 years into the
951. 33:35future, what does an iOS developer look
952. 33:37like?
953. 33:38Yeah, I've heard it described as kind of
954. 33:40the collapsing of hardware and software
955. 33:42where it's the hardware that's now making
956. 33:44the software and there's not a person in
957. 33:47between the two, which I think is an
958. 33:48interesting idea.
959. 33:49Um, I still think there's going to be
960. 33:52humans involved in that somewhere, but, uh
961. 33:54but it's interesting. interesting way to
962. 33:55think about it.
963. 33:56The final thing I will say is that I do
964. 33:59share the opinion of Anthropic CEO, Dario
965. 34:03Amode, that while all of this is scary
966. 34:06potentially dangerous, and there are so
967. 34:08many ethical problems, and the alignment
968. 34:11issue of artificial intelligence is a
969. 34:14real issue that unfortunately few
970. 34:16companies are thinking about.
971. 34:18But I do share his optimistic view from
972. 34:20his latest essay that I think it's called
973. 34:23the adolescence of AI. that through over
974. 34:28the long, or maybe not so long, arc of
975. 34:31human history, we as a species have been
976. 34:34surprisingly capable to adapt and find
977. 34:36new roles for ourselves through the many
978. 34:39industrial, well, not so many, the couple
979. 34:42of industrial revolutions that we've
980. 34:43had.
981. 34:44But through the arc of human history, we
982. 34:47have constantly replaced our previous
983. 34:50roles in any field and found new roles
984. 34:54for ourselves.
985. 34:54And I do think I do think we will
986. 34:58collectively as a society as a species, we
987. 35:01will come to an agreement in terms of
988. 35:03what's the role of humans.
989. 35:04I don't think we will end up, you know
990. 35:067000000000 people on earth.
991. 35:08I don't think we will end up altogether
992. 35:09on a beach somewhere doing nothing.
993. 35:12No, okay.
994. 35:12Yeah, and you've already, you've already
995. 35:14seen that.
996. 35:14I mean, the people who are squeezing
997. 35:16productivity out of AI, it's allowing
998. 35:18them to do things they wouldn't have had
999. 35:20time for anyway, right?
1000. 35:22before they wouldn't have had time to
1001. 35:24extend themselves and do all these
1002. 35:26different things.
1003. 35:26Now they're just doing more instead of
1004. 35:28less.
1005. 35:28I mean, I think the idea that is, I think
1006. 35:31that's called the Jevons Paradox, where
1007. 35:34despite improvements in a, in, in a
1008. 35:37technological field that was supposed to
1009. 35:39make something more efficient, it didn't
1010. 35:41make us work less, it made us work more.
1011. 35:43So I do believe that's gonna happen
1012. 35:47again.
1013. 35:48Hey, I've felt that myself.
1014. 35:50There are projects that I've taken on and
1015. 35:52done that I wouldn't have had time for
1016. 35:54you know, 6 months ago and now I can do
1017. 35:56it.
1018. 35:56Yeah, but to get to that point, we will
1019. 35:59need to have these conversations and, and
1020. 36:01I hope we will, we will continue to be
1021. 36:04able to have these conversations with
1022. 36:05this perspective of like, if AI is here
1023. 36:07to stay, what are we going to do?
1024. 36:11And we need to take a look at ourselves
1025. 36:12and we need to think of the ramifications
1026. 36:14both political, economical, and societal
1027. 36:17ramifications, right?
1028. 36:19When it comes to when it comes to the
1029. 36:21when it comes to the software that we use
1030. 36:23on our phones, maybe at some point in our
1031. 36:25glasses, on our computers.
1032. 36:27We need to think about those problems, so
1033. 36:29we need to think about our roles, right
1034. 36:31as people managing that.
1035. 36:33And so I don't think this episode has any
1036. 36:37major revelations, it's more of a, it's
1037. 36:40more of a conversation and maybe, I don't
1038. 36:43know, like a therapy session between me
1039. 36:45and you, but it's, uh, it's the kind of
1040. 36:47thing that's been on my mind.
1041. 36:48Like, if software is changing, how are
1042. 36:52people also changing in the process?
1043. 36:53Right.
1044. 36:54And it really goes to the heart of what
1045. 36:55AppStories has always been about.
1046. 36:57I mean, it's not just about what the apps
1047. 36:59do.
1048. 36:59It's the cultural and economic and
1049. 37:01societal implications of apps.
1050. 37:04It's really not been that long that we've
1051. 37:07lived in a world that was dominated by
1052. 37:09apps on a little slab of glass in our
1053. 37:11pocket.
1054. 37:12And now we've got something different
1055. 37:14changing now too.
1056. 37:16So, yeah, it's wild.
1057. 37:17Well, I think that's an episode of app
1058. 37:21stories.
1059. 37:21For this week, we're going to talk about
1060. 37:23Creator Studio in the pub show, for app
1061. 37:25stories plus members.
1062. 37:26I obviously wanna thank you, John, for
1063. 37:30this excellent conversation.
1064. 37:32You can find the both of us, just
1065. 37:34blogging away at the newly updated
1066. 37:36redesign, integrated, unified, MacStories
1067. 37:39net.
1068. 37:40You can find us on all the social media.
1069. 37:42You can find us on Mastodon, Blue Sky.
1070. 37:44And I really hope you check out the new
1071. 37:46MacStories with the club integration and
1072. 37:48AppStories under all under one roof.
1073. 37:51And we have a lot more new features
1074. 37:54coming soon.
1075. 37:54And, you know, until then, say goodbye
1076. 37:58John.
1077. 37:58Chop, there we go.