Vox

Leitura de 39 min

AI Revisited - part 2

Summary

In this episode of Rework, host Kimberly Rhodes talks with Jason Fried, CEO of 37signals, about his personal and professional use of AI. Jason shares how he uses tools like Claude and ChatGPT primarily as an editor to refine his writing, ensuring it aligns with customer language, and as a prototyping tool to quickly test product ideas without burdening his team.

Jason details specific examples, such as using AI to analyze customer testimonials for language patterns and to generate fake chat data for UI design in Basecamp 5. He emphasizes the value of AI in speeding up his workflow and avoiding the ‘owner’s word weighs a ton’ problem, where his requests carry undue weight and disrupt others’ work.

The conversation shifts to 37signals’ product strategy regarding AI. Jason explains that while competitors have heavily integrated AI, the company is taking a cautious, hybrid approach. They are exploring native AI features for Basecamp 5 while also preparing for a future where users bring their own AI agents into the product. The goal is to maintain simplicity and avoid building features that could quickly become obsolete.

Finally, they discuss the broader impact of AI on jobs and customer service. Jason argues that while AI can increase productivity, human support remains a crucial competitive advantage. He criticizes companies that use AI to create frustrating support labyrinths and reaffirms 37signals’ commitment to providing direct access to knowledgeable, long-term human support staff.

Recommendations

Tools

  • Claude — Jason uses Claude alongside ChatGPT for editing writing and prototyping product features, noting they have different ‘house styles’ he can leverage.
  • ChatGPT — Used in conjunction with Claude for editing and analysis tasks, such as aligning marketing language with customer testimonials.
  • OpenClaw — Mentioned as an example of the emerging trend of ‘always-on’ AI agents that can operate autonomously, signaling a future where users bring their own agents into software products.

Topic Timeline

  • 00:00:00 — Introduction and Jason’s daily AI use — Kimberly introduces the episode and asks Jason about his day-to-day use of AI. Jason explains he uses AI primarily as an editor for writing, using both Claude and ChatGPT to analyze customer language from testimonials and refine his own prose to better match how customers speak about Basecamp features.
  • 00:03:14 — AI for prototyping and product development — Jason describes using AI to prototype product features quickly, bypassing the need to ask other team members. He gives the example of prototyping a ‘recent chats’ feature for Basecamp 5’s sidebar, allowing him to test an idea instantly. He also mentions Chase using AI to generate fake PDFs with customer data for demos.
  • 00:07:22 — Generating fake data for UI design — Jason elaborates on using Claude to generate realistic, multi-user chat data over a three-week period for designing the Campfire UI. This allows him to see the interface with varied content without manually creating fake accounts and conversations, dramatically speeding up the design exploration process.
  • 00:10:35 — Evolution of AI adoption and skepticism — Kimberly asks if Jason, like David, has recently become more onboard with AI. Jason says he wasn’t initially skeptical but is now ‘sold’ on its utility, especially for product development tasks he would have previously had to ask others to do, appreciating the speed and autonomy it provides.
  • 00:12:19 — AI strategy for 37signals products — The discussion turns to AI integration in 37signals’ products. Jason explains they explored features in Fizzy but found some, like automated summaries, were ‘adequate but not interesting to read.’ He reveals the team is exploring avenues for AI in Basecamp 5, balancing native features with enabling users to bring their own AI agents.
  • 00:14:41 — The future of AI agents and product integration — Jason discusses the emerging trend of always-on AI agents (like OpenClaw) and predicts users will bring their own agents into products like Basecamp. He sees this as a shift that may reduce the need for heavy custom AI integration, allowing the company to focus on core product improvements while making it easier for external agents to access data.
  • 00:18:29 — Balancing polarized customer expectations on AI — Kimberly notes polarized customer feedback: some demand AI features, while others plead to keep Basecamp AI-free. Jason frames this as the classic software balance of adding power-user features without complicating the core experience. He believes they can satisfy both sides by offering some native AI assistance while allowing power users to integrate their own agents.
  • 00:21:51 — AI’s impact on jobs and company size — Kimberly asks about tech layoffs attributed to AI. Jason suggests companies often use AI as an excuse for layoffs driven by other pressures. He reiterates his belief that most companies are too large and that AI should boost productivity, not just replace people. He contrasts this with 37signals’ long-standing small-team philosophy.
  • 00:23:34 — The irreplaceable value of human customer support — They discuss the frustration of AI-only customer service. Jason argues that while AI can handle simple queries, human support is invaluable for nuanced issues, empathy, and de-escalation. He highlights 37signals’ competitive advantage in having highly trained, long-term support staff and commits to always offering a direct human support path.

Episode Info

  • Podcast: REWORK
  • Author: 37signals
  • Category: Business Business Management
  • Published: 2026-03-04T09:00:00Z
  • Duration: 00:28:01

References

Podcast Info

Transcript

[00:00:00] Welcome to Rework, a podcast by 37Signals about the better way to work and run

[00:00:03] your business. I’m your host, Kimberly Rhodes,

[00:00:06] joined this week by Jason Fried, CEO of 37Signals.

[00:00:11] Now we talked a couple of weeks with David about AI and this new energy

[00:00:15] around it, specifically how our programmers are using it internally.

[00:00:18] This week I thought we would talk with Jason about his thoughts on AI and

[00:00:23] from a product perspective, where our thoughts are. So Jason,

[00:00:26] before we talk about our products,

[00:00:27] let’s maybe talk about just AI in general.

[00:00:29] Are you using this on a day to day basis and what kind of use are you getting

[00:00:33] out of it so far?

[00:00:34] Yeah. You know, I use it for all sorts of different things on the personal

[00:00:37] side, plenty of stuff, business side.

[00:00:39] What I’ve been using it most for lately is actually as sort of an

[00:00:43] editor. So I do a lot of writing. So I’m currently writing,

[00:00:47] for example, a new basecamp.com homepage.

[00:00:50] And I wrote this piece as sort of a letter form,

[00:00:52] which is kind of how I approach these things.

[00:00:55] And I wanted to make sure that I was speaking really plainly and clearly.

[00:01:00] And I thought I was,

[00:01:01] but I also wanted to really match it up with a lot of the language that our

[00:01:04] customers use very specifically.

[00:01:06] So we had this page on our site,

[00:01:08] basecamp.com slash customers, which has about,

[00:01:10] I think it’s close to a thousand customer testimonials.

[00:01:14] And I use both Claude and ChatGPT.

[00:01:16] I kind of use them almost in competition with each other to kind of

[00:01:19] hone in on something because they each have their own style.

[00:01:21] And I don’t like their house styles necessarily,

[00:01:23] but they can kind of somehow help me get somewhere that I’m comfortable with in

[00:01:28] a way that I like.

[00:01:29] So what I did was I pointed them both at the customer testimonial page and

[00:01:33] just said like internalize the language our customers are using,

[00:01:36] how they describe things, what they call things. Because for me,

[00:01:39] I might call a feature to do’s and most people might call it tasks.

[00:01:43] Now I know what it’s to do is the tool and basecamp is called to do’s,

[00:01:45] but maybe people are calling it tasks and maybe we should call it tasks.

[00:01:48] But I want to make sure that what I wrote will land with more people and

[00:01:52] just land in their own mental model of what they’re thinking about and how

[00:01:55] they’re thinking about the product.

[00:01:56] So I asked it to internalize all the language and then read my letter and

[00:02:01] then make some suggestions for ways to tweak the language,

[00:02:05] not to tweak the letter or the tone,

[00:02:07] but make sure that I’m more aligned with how our real world customers

[00:02:11] speak about these things and call these things and name these things.

[00:02:15] So for example, I was using things like stay on top of things,

[00:02:19] make sure you know what’s going on. And I like those phrases,

[00:02:22] but our customers just say like organized. They just say like,

[00:02:25] I like that I’m organized. Basecamp keeps me organized.

[00:02:28] I’m so much more organized. So I don’t always take the suggestions from AI,

[00:02:33] but I think it’s a really good way to gut check and go, this phrase,

[00:02:36] this word is this lined up with what people are thinking in their head.

[00:02:40] So I’m not asking it to write the thing. I don’t like that,

[00:02:43] but I am asking it to see things that I can’t see and hold a lot of

[00:02:47] context in its head that I can’t hold and go, you know what?

[00:02:51] These four words would probably land better if they were these four words.

[00:02:54] And maybe I’ll use all four. Maybe I’ll use three, whatever it is.

[00:02:57] I also tend to sometimes write in a way where I have like one sentence too

[00:03:02] much in a paragraph, just like one too many sentences.

[00:03:05] And so I’ll often say like, turn this into three instead of four,

[00:03:08] but don’t just concatenate them. You know, like really what needs to go

[00:03:11] here. And so that’s the kind of stuff I’ve been using it for primarily.

[00:03:14] That said, I also have been digging into,

[00:03:17] we’re doing Basecamp 5 right now.

[00:03:18] So I’ve been digging into some features in the product and getting in there

[00:03:21] and having it tweak some things for me and change some things for me and

[00:03:23] quickly prototype some things for me.

[00:03:24] So I could see if this idea I have in my head is even worth pursuing.

[00:03:28] And it can throw some things together pretty quickly that are really,

[00:03:31] really handy that I could have done, but it would take a long time.

[00:03:34] And I don’t want to sink that kind of time in to find out if

[00:03:36] something’s worth doing in the first place.

[00:03:38] Okay. When you say prototype, what do you mean? Like,

[00:03:40] are you having it build something for you?

[00:03:42] Like that you can physically see?

[00:03:45] Yeah. So I’ll give you an example.

[00:03:47] We now have this feature in the product, but in Basecamp 5,

[00:03:50] there’s a sidebar with your pings in it. Currently in Basecamp 4,

[00:03:54] there’s a menu where you pull it down and you can see the pings are like

[00:03:57] direct messages. So you can pull the menu down and see,

[00:03:59] now we have it in the sidebar. If you want to open the sidebar,

[00:04:01] you can chat with people and not lose your place. What we do is we have these,

[00:04:05] currently we have little heads that show up,

[00:04:06] little avatars that show up when someone pings you.

[00:04:09] And if you don’t have the bar open,

[00:04:11] their head shows up with a little orange dot saying, Hey,

[00:04:12] there’s something new from Kimberly or something like that.

[00:04:14] That’s what that suggests. And if you hit P in your keyboard,

[00:04:17] it opens it up and you can start chatting. And that’s great, by the way,

[00:04:19] it’s great.

[00:04:20] But once I’ve clicked your avatar and I chat with you and I close the

[00:04:23] sidebar, like your face is gone. And so if I want to get back to you,

[00:04:26] it’s like an extra step or two to get back to that chat.

[00:04:29] And I found like I’m often talking to the same handful of people over the

[00:04:32] course of five hours or something like that.

[00:04:34] So it’d be nice to have like this menu build up of people I’ve talked to

[00:04:37] recently. We didn’t have this in the product though.

[00:04:40] We just had the new pings and I’d asked someone to build this,

[00:04:43] but they hadn’t gotten around to it yet because they were busy with something

[00:04:45] else. So I just built it. I mean, I didn’t do the building really.

[00:04:49] Claude did the building, but I asked it to like quickly prototype this

[00:04:54] idea. So heads would stick around after I chat with somebody.

[00:04:58] Now in the real product,

[00:05:00] I think we have like a six hour timeframe in which a head sticks

[00:05:03] around. If you don’t talk to that person again, it clears out.

[00:05:06] That way you’re not like having the sidebar full of people you’re not

[00:05:08] talking to, but it’s kind of an active culling of people.

[00:05:12] And I was able to quickly make this work for me.

[00:05:15] It probably wasn’t like robust enough to like ship in the product,

[00:05:18] but like it, it, I got it together in a way where it was working.

[00:05:22] Yeah. So you can see it.

[00:05:23] Yeah. I could mock it up myself, but that’s just looking at two states.

[00:05:28] I want to actually use it that way and see it work. And,

[00:05:31] and so like that’s something that I would have had to ask someone else to

[00:05:34] do. And I was able to do that myself. So I’m doing more of that.

[00:05:38] And that’s been really, really incredibly handy and like a breakthrough

[00:05:42] revelation kind of thing. Like, wow. And for me, it’s,

[00:05:45] it’s not so much like I can do things I couldn’t do before.

[00:05:47] Cause there’s certain things I could do before that I just chose not to do.

[00:05:51] So it’s mostly like,

[00:05:52] this is a huge speed up of time and I don’t need to bother somebody

[00:05:55] else.

[00:05:56] And I recognize how important this is because we’ve written this up in the

[00:05:59] past. Like a,

[00:06:00] I think it’s called the owner’s word weighs a ton or something like that

[00:06:03] was this old post I wrote up about how when you own the place like I do

[00:06:07] and David does, and if we ask someone to do something,

[00:06:10] there’s just more weight attached to that request,

[00:06:12] regardless of whether or not like we’re like, don’t worry about it.

[00:06:14] Don’t worry about it.

[00:06:15] Somehow there’s still more weight attached to that.

[00:06:17] So people tend to sort of drop what they’re doing to help.

[00:06:20] Yeah, of course.

[00:06:21] I really don’t want that most of the time,

[00:06:23] but it just comes with the territory.

[00:06:24] So it’s nice to not have to ask anybody and then not pull people off

[00:06:27] of things that even if I said, don’t worry about it,

[00:06:29] somehow they’d maybe get to it just cause they felt obligated in some

[00:06:32] way that I don’t want to put on them, but they still do.

[00:06:36] And so I just find that I’m able to bother people less and just get some

[00:06:38] stuff in my head quicker and decide if it’s even worth pursuing or not.

[00:06:42] Okay. Here’s a unique, I don’t know,

[00:06:45] unique to our use case situation that I came across with chat GPT

[00:06:49] specifically recently, we were doing a live base camp.

[00:06:53] Let me show you how this works,

[00:06:55] but there was data that we didn’t necessarily want to show his chase

[00:06:59] was actually doing this.

[00:06:59] And so he did like mockups of actual documents.

[00:07:03] He’s like, make me a PDF with this fake customer data,

[00:07:07] fake email addresses, fake phone numbers,

[00:07:09] so that we could show like, this is how it actually looks,

[00:07:12] but it’s not our account where you’re seeing these private things.

[00:07:15] And something as simple as that is like, okay,

[00:07:17] that just saved me from making it up.

[00:07:19] And it did in seconds.

[00:07:21] It’s huge.

[00:07:22] In fact, I did the same thing recently.

[00:07:24] I should have mentioned this example, Tim, glad he brought that up.

[00:07:26] So I’m running base camp locally, you know, as we’re developing it.

[00:07:28] So I have my own local database and I can screw around with it

[00:07:31] without messing up the real thing.

[00:07:33] Right.

[00:07:33] And so one of the things I’ll often do is I was working on a UI

[00:07:36] on campfire, our chat tool in base camp.

[00:07:39] I was working on the UI.

[00:07:40] I was curious, like, there’s a lot of stuff in there.

[00:07:42] I’d like to just get rid of it and move some stuff around or whatever.

[00:07:45] But the sample data, we typically have our seed data,

[00:07:48] which is like our default account that we use when we run things locally.

[00:07:52] Sure.

[00:07:52] Like our demo type account.

[00:07:54] Yeah, it didn’t have like a lot of chat history in it.

[00:07:58] And so I wanted to see a lot of different scenarios

[00:08:01] with file attachments and different people having conversations longer

[00:08:04] and shorter over many days and all sorts of different things going on.

[00:08:07] Right.

[00:08:08] And we just didn’t have that in the standard seed data in our basic data.

[00:08:11] Now, historically I would have asked maybe Marissa and she’s done this.

[00:08:16] And I think you worked with her on this too, right?

[00:08:18] This like incredible sample project stuff, which we have in the production

[00:08:22] version, but the local version, we didn’t have this and there was a scenario.

[00:08:24] I just wanted to sort of paint this scenario, but to make it happen

[00:08:28] was very complicated because it’s you have to like log in as multiple people

[00:08:32] to chat with fake multiple people.

[00:08:34] It’s you’ve been through it.

[00:08:35] You probably know.

[00:08:36] Yeah.

[00:08:36] So I asked Claude, I’m like, okay, just in this one campfire in this one

[00:08:41] project, give me like three weeks of conversations between five different

[00:08:45] users, including like file attachments and long conversations and short

[00:08:49] conversations and emojis and boosts and stuff, like just as much stuff

[00:08:53] as you can that would feel like a real conversation over three weeks,

[00:08:56] about five or six different topics, whatever, something like that.

[00:08:59] And I just said that and it populated the database and it’s like, it’s all done.

[00:09:03] I hit reload and it was pretty good, much better than nothing, not as good

[00:09:08] as what you and Marissa would have put together, but like I got it in

[00:09:12] three seconds or whatever, it was pretty instant at that point though.

[00:09:15] I realized like it was very repetitive and like too many one word responses.

[00:09:19] I’m like, eh, this doesn’t work, but let’s undo that.

[00:09:21] And then like, do it more like a, every response should be at least,

[00:09:24] you know, 12 words and maybe a couple that are really short and whatever

[00:09:28] thumbs ups and whatever.

[00:09:29] Anyway, after like a minute or two of screwing around, I just had a full

[00:09:34] campfire chat with fake real people with fake real conversations that

[00:09:40] helped me simply be able to design some ideas and see it in a

[00:09:44] bunch of different scenarios.

[00:09:45] And like that alone is so hugely helpful.

[00:09:49] It’s like what Chase was doing, but with chat, it’s especially

[00:09:52] hard because you have to come at it from multiple users.

[00:09:55] You don’t want to just chat with yourself that this doesn’t look like a real chat.

[00:09:58] So the fact that I was able to like spin up fake data like that, fake

[00:10:02] real data was so incredibly helpful.

[00:10:05] And then you could also just be like, okay, I’m done, undo it all.

[00:10:08] And it just like gets rid of it all.

[00:10:10] So it’s, it’s a sort of temporary fast worker to do some things

[00:10:15] just to get a sense.

[00:10:16] And then you can back out.

[00:10:17] You can also leave it there if you want, or just back out and do something

[00:10:20] else hugely helpful for me, especially as a designer, wanting to

[00:10:23] see things in a certain way.

[00:10:26] It’s hard to design when you don’t have that baseline data to design for.

[00:10:29] Yeah.

[00:10:30] And when I was talking to David about this, it sounded like he more recently

[00:10:35] not had come around AI, but as things have evolved more, he’s been like,

[00:10:39] okay, now I’m on board in a way that I wasn’t several months ago.

[00:10:43] Are you in that same boat where you’re more on board more recently, or

[00:10:48] have you always been pro AI in your like everyday use?

[00:10:52] I would say, and I won’t speak for David on this.

[00:10:56] I think where he was coming from was a bit of skepticism initially

[00:10:59] with code quality and the whole thing.

[00:11:02] Yeah.

[00:11:03] And so then I think he was impressed by the fact that it had evolved

[00:11:06] because back three, four months ago, it wasn’t as good as it is today.

[00:11:09] Right.

[00:11:10] Even just like the end of last year.

[00:11:11] Yeah.

[00:11:12] So I don’t think I was skeptical because I hadn’t done a lot of

[00:11:16] these things that I’m doing now.

[00:11:18] So I think I’m hopped on board.

[00:11:20] Let’s just say, I mean, I was using AI personally

[00:11:22] for all sorts of like just replacing Google stuff, right?

[00:11:25] You know, mostly.

[00:11:26] And then at work, I was using it for some writing stuff, but now getting

[00:11:29] into product development, I’m just impressed by how quickly I can do

[00:11:34] certain things that I would have had to ask someone for before, or like

[00:11:37] really get up to speed to do again.

[00:11:39] Yeah.

[00:11:39] So I’m not as, I wasn’t as skeptical and now I’m sold.

[00:11:43] I’m just like sold now.

[00:11:45] I think, you know, Claude could have done this six months ago too.

[00:11:48] I just wasn’t really like digging into it in that way, but I think I’ve

[00:11:51] just riding the wave right now.

[00:11:52] A lot of people, of course, are talking about it and showing off what it can do.

[00:11:55] I’m like, well, I should kind of get in there and figure this stuff out.

[00:11:57] And it’s been very, very helpful.

[00:11:59] I’m sure there’s a million ways.

[00:12:00] If someone was sitting next to me here who really knew all the things you

[00:12:03] could do and all the ways you could do it, I’m sure I could learn a lot

[00:12:06] more, but for now, the things I’m able to do, I’m really appreciative

[00:12:09] that I can do those things.

[00:12:10] I’m using it in the ways that I find it to be valuable for me.

[00:12:13] And I’m not like searching for use cases.

[00:12:16] I’ve got the things it’s doing for me right now.

[00:12:19] Okay.

[00:12:19] So now let’s talk a little bit about our products.

[00:12:22] Of course, we’re not going to reveal any spoilers, but tell me a little bit.

[00:12:26] We’re not, that would be more fun, wouldn’t it?

[00:12:28] I mean, it would, but I figured you didn’t want to.

[00:12:30] I don’t really have any spoilers at the moment.

[00:12:32] Tell me what you’re thinking in terms of AI and our products.

[00:12:37] Is that something that as we’re launching Basecamp 5, you’re digging

[00:12:39] into, is it a not now, and then also our other products, not just

[00:12:43] Basecamp, but Hey and Fizzy, kind of where are you thinking like the

[00:12:46] next step for AI is for us?

[00:12:48] Yeah.

[00:12:48] I mean, we explored it quite a bit in Fizzy, and I think there’s of course

[00:12:53] an endless amount of things you can do and maybe we should do at some point.

[00:12:56] Things we explored before in Fizzy weren’t entirely useful at the moment.

[00:13:00] That said, I can imagine, for example, duplicate detection.

[00:13:04] Like this bug looks like these three bugs.

[00:13:07] Is this the same thing?

[00:13:08] Things like that, I think would be handy.

[00:13:10] Yeah.

[00:13:10] We had some summary stuff, like summarizing a week of worth of work

[00:13:14] and what’s been going on, and we just found those to be like adequate,

[00:13:17] but like not interesting to read.

[00:13:19] And I think that that ultimately kind of pushed us back away from it for a minute.

[00:13:25] Like we can generate summaries, we can generate reviews of work, but like

[00:13:30] if no one really wants to read them, it didn’t feel right.

[00:13:33] So we kind of backed away from that.

[00:13:34] Anyway, that was a few months ago.

[00:13:36] David and I and Brian actually just caught up yesterday about AI and

[00:13:39] Basecamp 5, and we’re exploring some different avenues for inclusion

[00:13:44] of AI in the product.

[00:13:46] It’s also very, very interesting time because, and this is sort of David’s

[00:13:49] argument, which I buy, I also have other arguments that I’m trying to push

[00:13:53] forward that a lot of people have spent a lot of energy building a lot of

[00:13:58] custom AI stuff into their existing products, the alternatives in the

[00:14:01] market of competitors, alternatives, whatever other people in our sphere

[00:14:05] basically all have AI features in the products at some level and they

[00:14:08] talk about them very proudly on their sites.

[00:14:11] It’s very obvious that it’s like pervasive now and we don’t talk

[00:14:14] about it and we don’t have anything and that’s been an intentional decision

[00:14:18] so far and David’s point of view and I again, I agree with it and I also

[00:14:22] disagree with it in other ways, but I really agree fundamentally with it

[00:14:25] is that things have actually changed so much in the past month, like

[00:14:29] with OpenClaw, for example, and 24 seven running agents, things are

[00:14:33] just perpetually running for you and the ability for agents just to

[00:14:36] log in as normal people that people are going to end up bringing

[00:14:41] their own agents to our products.

[00:14:45] And just have them be normal users.

[00:14:47] Like just have your agent sign up for an account and then you can

[00:14:49] grant it access and you can bring all the knowledge it has.

[00:14:53] It can learn everything about your product in no time at all.

[00:14:56] We’ve seen this already.

[00:14:57] We already have agents in Basecamp in our Basecamp account.

[00:14:59] Like we’ve invited our own.

[00:15:01] And so you can see that there’s a lot of custom work you can do

[00:15:05] to try to do those things.

[00:15:07] Or maybe you can wait a little bit longer and the game’s going to

[00:15:11] change where OpenAI, Anthropic, Grok, Gemini, all these things

[00:15:17] will be offering these always on agents.

[00:15:20] Just like OpenClaw is, OpenClaw is like extremely technical right now.

[00:15:23] You got to set up your own server or use a virtual server.

[00:15:25] And it’s very complicated, but it’s a view into what things

[00:15:28] are going to be very, very soon.

[00:15:30] So what we can do is make the product simpler, clearer.

[00:15:35] We are working on some other stuff, some CLIs, command line

[00:15:38] interface for Basecamp that would make it easier for agents.

[00:15:40] So they don’t have to use a browser.

[00:15:41] We’re doing a bunch of that stuff too.

[00:15:43] But currently we’re thinking that people are going to be bringing a lot

[00:15:46] of their own stuff into Basecamp.

[00:15:48] However, there’s also stuff in Basecamp that we can be doing with AI that

[00:15:52] we’re going to be looking into doing for Basecamp 5 when we’ve already

[00:15:54] begun to think about these things.

[00:15:56] So I can’t reveal anything more than that, but I do think there’s

[00:15:58] going to be this hybrid world of some native AI features within a product.

[00:16:03] And then people are going to bring their own agents that are connected

[00:16:05] to all sorts of other things also, and that are going to know them

[00:16:08] really, really, really well.

[00:16:10] And they’re going to bring those into the product as well, just

[00:16:12] as if they were coworkers.

[00:16:14] You’ve probably seen Jeremy’s used doing this and David’s doing this

[00:16:16] and a few other people are doing this.

[00:16:17] And it’s, it’s incredibly impressive.

[00:16:19] And it’s interesting because we have to build anything to make this happen.

[00:16:21] And that’s, I think ultimately where the, the puck is going.

[00:16:24] That said, again, there are, I think, specialized things that we

[00:16:27] should be thinking about and things that are just more straightforward

[00:16:30] than having to think about signing for something somewhere else and then

[00:16:33] bringing it in versus just having a few things available to people.

[00:16:36] So it’s very exciting.

[00:16:37] It’s very much like on the edge of following what’s going on here and

[00:16:42] trying to determine what the best path forward is.

[00:16:44] And I’m actually very glad that we have not spent the past year building

[00:16:49] things that might be undone in a matter of weeks by a better way to do things.

[00:16:53] So, you know, there’s a point though, you can’t wait for someone

[00:16:56] else to invent the future that you want because you might have to wait

[00:17:00] 18 months and it might be too long.

[00:17:01] So like it’s figuring out the right timing for these things.

[00:17:03] It’s always challenging.

[00:17:04] It’s very hard to call the top of the bottom of any

[00:17:06] market or any situation.

[00:17:08] At some point though, you got to just go, oh, this is very interesting.

[00:17:11] This is going to change.

[00:17:11] What else can we do though?

[00:17:12] In the meantime, that still is very helpful.

[00:17:14] Yeah.

[00:17:15] I mean, it feels like you should always be early on the early side of

[00:17:18] things, but this is one situation where like, if you were early, you’re

[00:17:21] redoing basically what you’ve already done and with the small teams.

[00:17:24] Like we don’t have time to do that.

[00:17:26] Yeah.

[00:17:26] Look, we also have a small team.

[00:17:28] You know, we have 62 people in the company.

[00:17:30] A third of the company might be people who engineers and designers really,

[00:17:34] you know, ultimately, and like, AI is very helpful in, you know, allowing us

[00:17:37] to develop faster and with more fewer people and the whole thing, but

[00:17:40] ultimately we don’t have a company of a thousand people of which a hundred

[00:17:43] can go explore some of this stuff.

[00:17:45] So we still have to make decisions.

[00:17:46] There’s a lot of trade-offs to make about what’s worth focusing on.

[00:17:49] And we currently think that the core features of our product base camp,

[00:17:52] for example, that are ours that work the way we want them to work is a

[00:17:56] better place to invest most of our energy.

[00:17:58] We’ll still be investing some energy into AI and also have open arms and

[00:18:02] open doors for people to bring their own AI into base camp and make it easier

[00:18:07] for those AIs or for those agents to be able to access data in base camp.

[00:18:12] So we’re making that a lot easier.

[00:18:14] So I think we’re going absolutely the right direction here.

[00:18:16] And again, I’m glad we didn’t spend 10,000 engineer hours over the

[00:18:20] past year and a half or whatever it would be building stuff that’s like

[00:18:23] kind of going to be obsolete pretty soon.

[00:18:24] Yeah.

[00:18:25] It’s interesting because I see some of the customer feedback, people

[00:18:29] write in or comment on our YouTube videos.

[00:18:32] That people are very polarized by this topic.

[00:18:36] I’ve said it before, but there’s people who are like, when is base camp getting AI?

[00:18:39] Like you guys are late and there’s people who are like, don’t touch my base camp.

[00:18:42] It’s like the only piece of software I have that hasn’t been sucked in with shitty AI.

[00:18:47] So it is one of those like hard to balance between those different opinions

[00:18:51] and doing what’s right for the product.

[00:18:53] For sure.

[00:18:53] And we want to make both those sides happy.

[00:18:56] And I think we can, I absolutely think we can.

[00:18:58] Our approach is always to be as straightforward as possible.

[00:19:01] As no nonsense as possible.

[00:19:04] So like slathering AI on everything, everywhere, all the time is not going

[00:19:09] to be our approach and I’ve used tools that are like that now where

[00:19:12] everything is like AI first.

[00:19:14] And I just think it’s a bit of a novelty at the moment.

[00:19:17] And I think it’s wearing thin in some ways and you just have to be careful

[00:19:20] not to sort of ruin things because not everybody wants that all the time.

[00:19:23] It should be available for sure.

[00:19:25] And others are very gung-ho about it, want it everywhere.

[00:19:28] So the good news is they’ll be able to bring theirs and have it

[00:19:30] wherever they want and do whatever they want.

[00:19:33] Meanwhile, for those who don’t want to go down that road or aren’t

[00:19:36] sophisticated enough or uninterested enough in doing that, we have to

[00:19:39] provide some assistance for them as well, and give them some leverage

[00:19:43] that they didn’t have before and show them that this stuff is very

[00:19:45] powerful and very useful, but also not in your way.

[00:19:48] So it’s a delicate balance, but this is the same delicate balance

[00:19:51] we’ve been balancing on since the beginning of Basecamp.

[00:19:54] Everybody wants more stuff.

[00:19:56] This is the nature of software.

[00:19:57] Everyone wants more stuff and everyone has their two or three requests

[00:20:00] and they can’t understand why you haven’t done them yet.

[00:20:02] And then you’re like, well, there’s 85,000 people asking

[00:20:05] for two or three things.

[00:20:06] And some of those things overlap and many of them don’t.

[00:20:08] And then you also, you know, people will also say like, I love

[00:20:10] Basecamp because it’s so straightforward and simple and thank God

[00:20:13] because everything else is a mess.

[00:20:14] And you’re like, big reason for that is because we’ve, we’ve held

[00:20:16] back doing certain things that everyone’s been asking us to do.

[00:20:19] And it’s always a delicate balance.

[00:20:21] So it’s no different.

[00:20:22] Expectations are no different.

[00:20:23] They’re just about different things.

[00:20:24] So it’s always the thing we’ve been, we’ve been really good at,

[00:20:26] I think, which is understanding the limits, making things very accessible

[00:20:30] for a huge swath of people and not getting ahead of ourselves and

[00:20:34] making things complicated to benefit a handful of people who really

[00:20:37] want the most and the many.

[00:20:39] And instead just kind of figure out like what’s the right collection

[00:20:42] of things that makes the most sense for the most people that’s easy

[00:20:45] and approachable and understandable for nearly everybody.

[00:20:48] And also there’s some more power around the corner.

[00:20:51] If you really know how to get at it, you can do that.

[00:20:53] Like a good example of this.

[00:20:54] Well, let’s call Basecamp three.

[00:20:56] Actually all the way back to two kind of had a handful of tools that were,

[00:20:59] you know, you could add to do’s and messages and files and documents and

[00:21:03] schedule stuff basically in a project.

[00:21:06] And that was the same up until a couple of years ago, we added this

[00:21:11] feature to allow you to add multiple tools to a project.

[00:21:13] So you could have multiple to do sections.

[00:21:15] You could always have multiple to do lists, but I give two, two

[00:21:18] sections or multiple message boards, two separate chat tools, right?

[00:21:22] And so it’s still simple for everybody.

[00:21:24] Like there’s a collection of simple, straightforward tools that

[00:21:27] makes sense to everybody, but those who want to reach around the corner

[00:21:30] and pick out a few more things off the shelf and put it on their project, they

[00:21:34] can, but it never gets in somebody’s way if they don’t want to think about

[00:21:38] the fact that they can do that.

[00:21:40] That’s always the line we’re trying to tow here.

[00:21:42] And so I think the same thing will be true with AI.

[00:21:44] Okay.

[00:21:45] A little off subject, but I’ve read a lot or heard recently about the whole

[00:21:51] tech surge of people being laid off because AI is going to take over all

[00:21:55] these jobs and we don’t need programmers and we don’t need all of these

[00:21:58] different positions.

[00:21:59] I kind of wanted to get your take on that.

[00:22:00] One of the things I think of in particular that drives me crazy with

[00:22:04] companies, their use of AI is supports and offloading that support to AI

[00:22:10] robots where you can’t talk to a real person.

[00:22:12] I want to kind of get your take on the company’s philosophy when it

[00:22:16] comes to the roles that we have here and AI, how those might be supplemented

[00:22:22] or not.

[00:22:23] My longstanding opinion about most companies is they tend to have too

[00:22:26] many people.

[00:22:27] I think companies tend to be too big, teams are too large, and we’ve

[00:22:31] intentionally kept our company as small as we possibly can, like since

[00:22:34] we existed, we got a little bit bigger than we are today, but you

[00:22:38] know, we’re about 60 people.

[00:22:40] That feels like really good.

[00:22:41] And a lot of the people in our industry have teams of hundreds or thousands of

[00:22:46] people on their company.

[00:22:47] And I’ve just never really understood that.

[00:22:49] Fair enough, whatever.

[00:22:50] So in general, I think that companies will often look for reasons to let

[00:22:55] people go that may or may not be true.

[00:22:58] At some point they lay, you know, when their numbers aren’t right and

[00:23:01] Wall Street, if they’re public, Wall Street’s demanding this or

[00:23:03] demanding that, it’s easy to lay people off and you can say, we’re

[00:23:07] laying people off because AI is going to make us more efficient or more

[00:23:10] productive.

[00:23:11] And that might be true, but also just might be an excuse to lay people off.

[00:23:15] So I don’t know, I don’t know.

[00:23:17] Sure.

[00:23:17] There’s no question that it should make people more productive, but I

[00:23:21] think people should have probably been a lot more productive to begin with.

[00:23:23] I don’t think you need a team of 12 people working on something

[00:23:26] that’s relatively small, you know?

[00:23:28] So it’s all like one of the same for me there.

[00:23:30] I agree with you that like in some cases, AI based customer service is

[00:23:34] actually quite good if you have a very simple, straightforward question.

[00:23:37] It’s also incredibly frustrating when you have something a little bit more

[00:23:41] nuanced and you just have this boiling urge, like I just want to talk to

[00:23:47] somebody who will understand where I’m coming from, who’s not as

[00:23:52] intelligent as AI technically, but totally gets what I’m talking about

[00:23:55] because they just, they’re human and they get it right.

[00:23:58] Now there’s also terrible human customer service out there as well.

[00:24:02] And that’s because people aren’t trained well and companies typically

[00:24:05] see it as a cost center, so they try to find the lowest common denominator.

[00:24:08] So it’s not that humans are better or worse at this.

[00:24:11] It’s like you want highly trained long-term humans who know a

[00:24:15] product inside and out are extraordinarily good and that’s what we offer.

[00:24:21] We do offer some AI help in some places.

[00:24:26] You hit the little question mark and you can ask a question and stuff,

[00:24:29] but you just email support at basecamp.com and you’re getting a human.

[00:24:32] Right.

[00:24:32] So if we’re around in 245 years and we’re all dead and that like might be

[00:24:37] different in the short term near term here, I think humans are hugely important.

[00:24:41] I think we have a massive advantage because we have incredibly good humans

[00:24:45] on our support team, many of which have been here for many, many, many

[00:24:48] years, some of which been here for 15 years on support.

[00:24:52] Like this is a career job here, not just a temporary job, which it is

[00:24:56] in most places, kind of a temp job.

[00:24:59] And we’ve developed incredibly good people who have a huge depth of

[00:25:02] knowledge about how our products work and they really, really care.

[00:25:04] And I think it’s a massive competitive advantage for us.

[00:25:06] And I would not want to give that up.

[00:25:08] That said, there’s also times you just want to get a quick answer.

[00:25:10] So we should have both and make both available, but it shouldn’t feel like

[00:25:13] you must go through the AI and then get pissed off enough to get a human.

[00:25:18] I don’t ever want to have us do that.

[00:25:20] Those are the experiences I never want to have with other people’s companies.

[00:25:23] And by the way, it’s so different than like old school support,

[00:25:25] like phone trees, you know, like, oh my God, like press five.

[00:25:29] And then the recording is so slow.

[00:25:30] We’re like, and you just like slam the zero button or whatever, and try to

[00:25:34] like break through it, you know, I don’t want anyone to ever feel that like

[00:25:38] that with us.

[00:25:39] Yeah, I, the reason I brought it up is I recently had this terrible

[00:25:43] experience with an airline, trying to get a question answered and talking

[00:25:47] to their AI bot, trying to get a real person like, okay, you’re not

[00:25:51] answering my question, so I need to get to a real person and literally

[00:25:55] being in a circle of no, I’m the bot and I can help you.

[00:25:58] Well, you’re not helping me.

[00:25:59] No, I’m the bot.

[00:26:00] I can help you.

[00:26:00] I’m like, oh my gosh, infuriating.

[00:26:04] Well, there’s this sort of know-it-all complex.

[00:26:07] And you’d imagine like AI technically does know more than any human at

[00:26:11] this point, essentially, but it’s still different because people relate.

[00:26:15] It’s not just about knowledge.

[00:26:17] It’s about relating to somebody and inventing, frankly, to a human who

[00:26:22] understands that you’re pissed and can like feel that and understand how to

[00:26:27] respond in a way that’s still just human to human, like humans want to

[00:26:30] talk to humans, especially when they’re pissed, they just do.

[00:26:34] I mean, you could also yell at an AI and like let it all out, not worry

[00:26:37] about insulting anybody either, but I don’t think that’s what people

[00:26:40] actually want.

[00:26:40] I think when people really get a little bit frustrated and sometimes

[00:26:43] when they’re writing support, they are, they are, and it might be

[00:26:46] because they had a bad day.

[00:26:47] It’s not that there’s the product is terrible.

[00:26:49] They may have had a bad day or who knows what happened, right?

[00:26:52] They’re, they have a deadline coming up and they can’t find this thing

[00:26:54] they knew was there and like, people are frustrated for all sorts of reasons.

[00:26:58] You want someone on the other side who can meet that, absorb it, understand

[00:27:02] it, and know how to work with you on it.

[00:27:05] Now it’s not that AI can’t be trained to do that.

[00:27:08] And I’m sure there’s some very sophisticated models that can do that,

[00:27:12] but that’s a technologist’s point of view.

[00:27:15] If I’m a human, there’s a point and it’s not a very deep point

[00:27:19] where I want to talk to somebody.

[00:27:20] I just still do maybe in 20 years.

[00:27:22] I don’t today.

[00:27:23] I do.

[00:27:24] I believe that’s true.

[00:27:25] I talked to our customers.

[00:27:26] I know it’s true.

[00:27:28] And we want to make sure that we’re never skimping on that.

[00:27:31] Yeah.

[00:27:31] Well, that seems like a perfect place to wrap it up.

[00:27:33] We will look forward to seeing what comes in the way as of AI,

[00:27:38] as we launch new products.

[00:27:40] This is a production of rework.

[00:27:41] You can find show notes and transcripts on our website at

[00:27:43] 37signals.com slash podcast, full video episodes on YouTube.

[00:27:47] And if you have a question for Jason or David about a better way to

[00:27:49] work and run your business, leave us a voice recording.

[00:27:52] You can do that at 37signals.com slash podcast question, or send

[00:27:55] us an email to rework at 37signals.com.

Visão de gráfico