Sell APIs and libraries for tools that make genai reliable, safe and predictable

Any app idea that you think you should build, just build a library or api for it instead. APIs and libraries will be picked up by vibecoding tools as well, instant distribution. Claude will do distribution for you :)

Someone else built the crowd, you just position yourself where the crowd already flows.

In software, this happens whenever you build on top of, inside, or alongside a platform that already aggregates user intent.

One way : designing an api or library

If your library becomes a dependency for popular projects, you get passive distribution through other people's code.

@impactology

Hmm, I think you are overlooking some opportunities for marketing here.

Just one example: If your product solves problems for users of another type of software, that the makers of the other software don't want to take on, they can be a source of referrals to your shop. One example might be a password manager referring people to antivirus / antimalware.

Often that physical placement of the storefront has planning behind it, too.

@impactology

I feel like I probably have, at least in terms of newsletter-style emails, or maybe breach notifications? But no, I don't have an example handy.

Another example that comes to mind is Taiga and PenPot. Different products by the same company https://blog.kaleidos.net/Taiga-and-Penpot-lead-our-new-chapter-at-Kaleidos/ In my experience they do less cross-promotion than they probably could. I am just a dabbler in using both products. They may do more cross-promo than I've seen, or have good reasons for not doing more.

What this means is someone else must have done the work to make the space/street popular or useful and simply by being in proximity to that shop I get to have visitors.

Where in software do you see this? Maybe plugin ecosystems of really popular software but what else

Vada pav seller gets to make money simply by being near a college and offices. Simply by being in the right shared space near people who need it most or most likely to buy it. No marketing needed.

Whereas for digital products or services you always have to market.

One huge advantage that a physical product based business has over software in terms of marketing is that a physical storefront can be in a good location and get footfall simply because it is in a shared space beside another shop the person could be browsing

Like you could be walking down the street to buy a shoe and come across a shop down the block selling pants

In software spaces are not shared, they are discrete. You have to make the person land on your landing page.

You can talk about this stuff without shame after being successful because money gives you confidence to be vulnerable

Because then failure can be called as a lesson, an iteration, data

4) Creating a working professional adult identity from scratch without any help rather than inheriting or borrowing it, that would be the theme of the podcast

5) Of finding your footing after being lost, all by yourself. Of reintegrating into society after being isolated. That could be another theme

A friend asked me in future once you become successful and asked to do a podcast interview what themes you'd cover

I said

1) Overcoming the struggles of having a predictable cash flow via tech, design work

2) Of creating and anchoring a professional identity without institutional affiliation, of signalling trust in one's skills via projects to pay you without it

3) Of navigating masculinity late, Of figuring out how to be social

The least that an LLM interface should have to even qualify it's designed, is to make the user know it's failing predictably

Like if you assume its going to fuck up, atleast have it limited to fucking up in only xyz ways so the user has a way to ration their time and energy to steer it.

Designing predictable failure modes in LLMs

One exception to tech that violates predictable comfort : musical instruments

It has managed unpredictability. For example a jazz instrument is unpredictable in output, but its control surface : keys, strings is highly legible.

Even there the uncertainty is contained, not chaotic.

Engineering can make a system work, but design is what makes it work for a human, within their cognitive bandwidth, emotional tolerance and need for predictability.

If a technology and it's control surface is so unpredictable that using it every time gives you anxiety over when it will break or what it will break it has failed to afford the basic use of technology : predictable comfort

Everything that affords human control over an activity predictably is design, like that which makes it possible for human mind and body, within its comfortable limits to take action is design's domain I think, not engineering or sciences.

The latter via the built environment, objects, technology help create a repeatable function but it's design that helps humans to control it predictably.

Which is why I think LLMs haven't been designed yet, only engineered.

RE: https://mastodon.social/@impactology/116353480943373841

This guy does a better job of explaining swe than multiple semesters of indian cs edu (our syllabi are outdated and our pedagogies suck, they have theory memorization based written exams, not project oriented)

https://youtube.com/playlist?list=PLui3EUkuMTPgZcV0QhQrOcwMPcBCcd_Q1&si=ArftSCTuX35j6UUM

A software developer or product design freelancer can't pitch their services as "you'll recoup your investment in my service by xyz time or outcome" that a gtm sales freelance guy still can, but then the gtm guy can't promise you sales without having a quality product built in the first place

you can TELL that this technology REALLY WORKS by how the people that made it and presumably know how to use it the best out of everyone CANT EVEN USE IT TO EDIT A FUCKING FILE RELIABLY and have to resort to multiple stern allcaps reminders to the robot that "you must not change the fucking header metadata you scoundrel" which for the rest of ALL OF COMPUTING is not even an afterthought because literally all it requires is "split the first line off and don't change that one" because ALL OF THE REST OF COMPUTING can make use of the power of INTEGERS.

@aredridel @mnl right, yeah. the thing i both love and find to be extremely sad is that one of the biggest things that LLMs do, especially for amateurs and hobbyists, is translate natural language into a set of calls to existing tools. it's awesome that people who wouldn't know what to look for can do that, but it's a real shame when it gets presented as "look what i made" rather than "look at what everyone else as made and isn't it amazing that i can just use it for free"

not saying that's what's happening here, speaking more generally from what i've seen with people who have never programmed before picking up the tool, and instead of being introduced to a community of open source devs who have been working to make things for free for everyone for decades, the person is only exposed to the oracle and that remains the only point of contact or exposure.