Kyle Zantos
Product designer and builder of AI tools and workflows, including a content system for Dive Club; Dive Club's first three-time guest.
Kyle Zantos did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest GitHub
What they believe
-
Ask agents for the closest possible representation of a thing, so an idea is approved on more than a best guess.
my directive is always give me the closest possible representation of the thing so that when I say, "Yeah, that's a good idea. Implement that." I'm doing it from a not best guess, but like I can tell that this is the direction we want to do.
youtube.comSaid 11 Aug 2026This on korrents
The bold line is korrents’ wording; the quote is theirs. These one: what they wrote before what they said aloud, one per source, newest first.
Feed
As its own page →Hiding
29 September
22 September
15 August
14 August
11 August
From one piece Kyle Zantos - Our AI design workflows have changed (again) 12 beliefs · youtube.com
-
korrents.com
If you are primarily designing something, it makes all the sense in the world to own the front end.Their words
If I am primarily designing something, like it makes all the sense in the world to me to own the front end. Of course, if I'm the one designing it, why wouldn't I own that? That it's insane. I want to play the songs that I write, you know?
-
Their words
there was a a period of time that being early to the AI game was an attribute to itself that is becoming less and less true already. Like so many people have caught up and it doesn't take all that.
-
korrents.com
Going wide cheaply with HTML, then making the real app once there is a sketch of it, is so much better.Their words
and going wide now cheaply with HTML stuff and then getting into now let's make the real app because we actually kind of have a sketch of what the app will look like is is so much better.
+ 9 more
-
Their words
like the ideation happened like in v 0ero on top of like a full react web app which is like now just like laughable. It's like that's so inefficient.
-
Their words
I mean one of the things was building a lot of custom tooling. The HTML shortcut like takes care of so much of that. Like you said the the show me the options and approved and I discuss like that whole area would have been a tool in your case was a tool for a minute uh that now we don't have to do.
-
Their words
We're continuously trying to describe and like put something into the agents hands quote unquote or or make it something they can touch so that the touch identifier of exactly this thing and you can see all the code and what it is that I'm talking about and my talking about it that combo is seems like a theme in a lot of the stuff we've been building.
-
Their words
I think you keep reminding me I need to not assume how good it might be or might not be at this thing that I want it to do.
-
Their words
I'm just thinking differently of like yeah it's like an eval. It's like a visual eval that us not being in the code, we can make more complex tooling and flows like this and autonomous runs that we still get to manifest and tweak and control basically the same way that we're used to.
-
korrents.com
Seeing each version an agent pipeline produces, instead of a black box, shows where the process is too weak.Their words
instead of having the black box of like I told you to go through all these iterations now I see V3, but I don't know what happened along the way. And seeing it visually like this, especially while I'm developing this pipeline, lets me find like, man, review between one and two, the review is too weak.
-
Their words
I didn't want to keep telling it to fix stuff like you would tell like an intern. Let's hit the bar of solid and then I'm working from what I like and don't like, not just like this is an obvious poor visual design decision.
-
Their words
So just visualizing everything from your agents has been huge.
-
korrents.com
Ask agents for the closest possible representation of a thing, so an idea is approved on more than a best guess.Their words
my directive is always give me the closest possible representation of the thing so that when I say, "Yeah, that's a good idea. Implement that." I'm doing it from a not best guess, but like I can tell that this is the direction we want to do.
23 July
9 July
30 June
30 May
- design-motion-principles — A motion design skill with two modes — build interactive components with purposeful motion, or audit existing animations. Context-aware, per-designer guidance distilled fro…
- thumb-first — Mobile reviews need two reads: taste and device reality. Runs design judgment plus PWA/platform checks, then merges them into one prioritized report.
6 May
22 April
16 April
5 April
4 April
3 April
Nothing matches.
Beliefs
Korrents What they believe 12 beliefs — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
Ask agents for the closest possible representation of a thing, so an idea is approved on more than a best guess.
my directive is always give me the closest possible representation of the thing so that when I say, "Yeah, that's a good idea. Implement that." I'm doing it from a not best guess, but like I can tell that this is the direction we want to do.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Visualising everything your agents produce has been huge.
So just visualizing everything from your agents has been huge.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
An agent should reach a solid bar on its own, so the designer works from what they like and don't like rather than correcting obvious poor visual design like an intern's.
I didn't want to keep telling it to fix stuff like you would tell like an intern. Let's hit the bar of solid and then I'm working from what I like and don't like, not just like this is an obvious poor visual design decision.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Show 9 more
Seeing each version an agent pipeline produces, instead of a black box, shows where the process is too weak.
instead of having the black box of like I told you to go through all these iterations now I see V3, but I don't know what happened along the way. And seeing it visually like this, especially while I'm developing this pipeline, lets me find like, man, review between one and two, the review is too weak.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
A visual eval lets designers who are not in the code build complex tooling and autonomous runs they can still shape and control the way they are used to.
I'm just thinking differently of like yeah it's like an eval. It's like a visual eval that us not being in the code, we can make more complex tooling and flows like this and autonomous runs that we still get to manifest and tweak and control basically the same way that we're used to.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Do not assume how good a model will or will not be at the thing you want it to do.
I think you keep reminding me I need to not assume how good it might be or might not be at this thing that I want it to do.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Giving agents something they can touch, together with your spoken description of it, is a recurring theme in the AI design tooling being built now.
We're continuously trying to describe and like put something into the agents hands quote unquote or or make it something they can touch so that the touch identifier of exactly this thing and you can see all the code and what it is that I'm talking about and my talking about it that combo is seems like a theme in a lot of the stuff we've been building.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
The HTML shortcut takes care of much of the custom tooling designers used to build.
I mean one of the things was building a lot of custom tooling. The HTML shortcut like takes care of so much of that. Like you said the the show me the options and approved and I discuss like that whole area would have been a tool in your case was a tool for a minute uh that now we don't have to do.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Ideating in v0 on top of a full React web app now looks laughable and inefficient.
like the ideation happened like in v 0ero on top of like a full react web app which is like now just like laughable. It's like that's so inefficient.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Going wide cheaply with HTML, then making the real app once there is a sketch of it, is so much better.
and going wide now cheaply with HTML stuff and then getting into now let's make the real app because we actually kind of have a sketch of what the app will look like is is so much better.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
Being early to AI was once an attribute in itself, and that is becoming less and less true as so many people catch up.
there was a a period of time that being early to the AI game was an attribute to itself that is becoming less and less true already. Like so many people have caught up and it doesn't take all that.
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
If you are primarily designing something, it makes all the sense in the world to own the front end.
If I am primarily designing something, like it makes all the sense in the world to me to own the front end. Of course, if I'm the one designing it, why wouldn't I own that? That it's insane. I want to play the songs that I write, you know?
Kyle Zantos - Our AI design workflows have changed (again) Said 11 Aug 2026
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
Where they publish
Site zantos.co His portfolio: selected work, skills and tools, and what he is building now.
Projects he designed and built (an AI content tool for Dive Club, a personal agent called Higgins, the Headroom Mac menu bar app for Claude and Codex limits), a case study from 2017 to 2021, and small apps made for family and friends.
Link verified 4 Oct 2026.
x.com @kylezantos His X account.
Linked from his portfolio's social icons; the feed itself was not read when this was verified.
Link verified 4 Oct 2026.
Interviews As a guest Long-form interviews on other people’s channels, chosen and not exhaustive.
His own voice by his own participation: Dive Club's episode on HTML artifacts as a design workflow, node-based pipelines for variations, video feedback to agents and where design is heading. His third appearance on the show.
Link verified 4 Oct 2026.
GitHub kylezantos His open code: small tools for working with AI coding agents and Mac apps.
Public repos include a Raycast media compressor, a Claude Code skill for delegating implementation from Fable to Opus subagents, a Tart VM runner for Xcode builds and a Homebrew tap for his tools.
Recent
- media-compressor-raycast — Raycast extension for compressing images and videos with full control over quality, codec, and size settings. 29 Sept 2026
- New repoheadroom-downloads — Public signed Headroom installers and updater files. App source is maintained separately. 22 Sept 2026
- loom-walkthrough-brief 15 Aug 2026
Show 12 more
- fable-opus-delegate — Claude Code skill for Fable sessions: delegate implementation to Opus subagents, Fable stays orchestrator + QC 14 Aug 2026
- homebrew-tap — Homebrew formulae for Kyle Zantos's tools 23 Jul 2026
- finding-language-market-fit — Language-Market Fit copy skill — compose or audit copy grounded in real customer language (Matt Lerner framework) 9 Jul 2026
- skill-distillery — Create or audit AI agent skills. Synthesizes source material bottom-up into strategic, interactive skills. Works with Claude Code, Codex, OpenCode, and 35+ agents. 30 Jun 2026
- design-motion-principles — A motion design skill with two modes — build interactive components with purposeful motion, or audit existing animations. Context-aware, per-designer guidance distilled fro… 30 May 2026
- thumb-first — Mobile reviews need two reads: taste and device reality. Runs design judgment plus PWA/platform checks, then merges them into one prioritized report. 30 May 2026
- skill-picker — Descriptions lie sometimes. Deep-reads installed SKILL.md files, compares close candidates, and picks the skill chain that actually fits the task. 6 May 2026
- convex-setup-skill — Convex setup without the mystery gap. Detects your app, guides database setup, and adds Clerk auth only when it belongs. 22 Apr 2026
- intent-engineering — Before the polish pass, ask what the page should make people do, notice, and feel. Start or audit a design from bedrock intent. 16 Apr 2026
- mtg-arena — MCP server that gives Claude Code access to your MTG Arena cards, decks, collection, and inventory 5 Apr 2026
- responsive-craft — Responsive work breaks in the in-between sizes. Build, retrofit, or preview layouts across sticky coordination, scroll regions, tables, and dashboards. 4 Apr 2026
- compounding-efforts-life — Sort the work that compounds from the work that only disappears when you finish it. Quick triage plus weekly, monthly, and quarterly planning. 3 Apr 2026
Link verified 4 Oct 2026. Recent items update automatically from the channel.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.