The argument in one line.
Design.md lets non-designers attach a single portable design system file to AI prompts, maintaining consistent visual identity across web, mobile, slides, and motion instead of each output drifting into generic mediocrity.
Read if. Skip if.
READ IF YOU ARE…
- A founder or solo maker building a web or mobile product who lacks design experience and wants visual consistency across multiple formats without hiring a designer.
- A designer working with AI tools or agents who needs a portable system file to attach to prompts so output stays coherent across landing pages, mobile, and motion.
- Someone who ships fast and currently accepts generic or inconsistent design as a tradeoff, but wants to elevate visual quality without slowing down iteration cycles.
SKIP IF…
- You're already comfortable writing design specs or briefs in your current format — this specifically teaches markdown-based systems, which may feel like unnecessary relearning.
- Your work is primarily fiction, brand strategy, or print-focused rather than digital product and web — DESIGN.md is built for screen-based, multi-format outputs.
The full version, fast.
Design.md is Google's open-sourced markdown spec that captures a design system's typography, color, spacing, components, and WebGL into a single portable file you attach to any AI prompt, solving the design drift that makes one-shot landing pages collapse into generic output by page two. The workflow pairs a design.md with an HTML reference and composable 'skills' � reusable prompt fragments for effects like lasers, skeuomorphism, or copy polish � then iterates through hundreds of remixes across web, mobile, slides, and motion while keeping the same DNA. Treat the file as portable design memory between Cursor, Codex, Aura, or Stitch, build taste by studying niche references obsessively, and accept that beautiful output still requires roughly a thousand judgment calls per product, not one prompt.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:56
01 · Greg's cold open
Greg promises that by the end of the episode 'you will become a designer' — DESIGN.md is the unlock, his friend Meng will demo the workflow.
00:56 – 01:07
02 · Sippin' Time bumper
Branded juice-box transition cuts from cold open to the conversation proper.
01:07 – 02:58
03 · What is DESIGN.md (Meng's framing)
Meng explains DESIGN.md as a portable blueprint that lets one beautiful design propagate across landing pages, mobile, slides, and motion — solving the 'great first screen, generic everything after' problem.
02:58 – 04:00
04 · Meng's prep + thank you
Meng credits Greg's pod for taking him from $3K to $15K MRR, then announces he prepared slides via Reply Slides / Hyperframes / Remotion. Sets up the screen-share.
04:00 – 06:00
05 · Five-point agenda + DESIGN.md = soul of design
Meng frames DESIGN.md as 'the soul of a design ported into the agent' — colors, typography, the system — same family as agents.md / skills.md / soul.md, but for visual design.
06:00 – 07:00
06 · Idea Browser ad break
Greg drops a mid-roll for his Thursday workshop on building businesses in an AI world via ideabrowser.com.
07:00 – 08:30
07 · One DNA, many mediums
Meng shows a slide and a promo video built from the same DESIGN.md — 'same DNA, different approach' — and points at a community where these markdown files live as free downloads.
08:30 – 10:00
08 · Recipe vs dish vs ingredients
Core metaphor: HTML is the finished dish, DESIGN.md is the recipe, skills are the ingredients. Combine all three to get repeatable beautiful output.
10:00 – 11:00
09 · Design Drift — the real problem
Meng names the failure mode: 'design drift.' You one-shot a great first screen, then every later page diverges. DESIGN.md is the cure because the system travels with the prompt.
11:00 – 13:20
10 · Why templates fail vs blueprints win
Templates lock you behind Figma/Framer/Webflow pixels — you can't tell an agent what they mean. DESIGN.md communities share the blueprint instead, so the AI gets a foundational system rather than a literal copy.
13:20 – 15:20
11 · Cookie-cutter is dead
Greg + Meng riff on Shopify/WordPress/Framer sites all looking identical — like a downtown core with homogeneous buildings. Purple-gradient era is over.
15:20 – 17:00
12 · Taste as the real moat
Meng reframes intelligence as 'making ultra-quick decisions about taste.' DESIGN.md is design memory you can transfer between Lovable, Cursor, Codex, Claude Code, Figma — your taste, portable.
17:00 – 18:50
13 · Local is the moat
Meng's products generate MD files in nested folders he can open in Codex/OpenClaude — 'because it's local' the agent can read the whole knowledge graph in one shot. Tokens, lock-in, and remote-only tools are the contrast.
18:50 – 20:40
14 · Demo setup — Aura, Variant, remix
Meng walks Variant.com and his own Aura — remix-click-remix loops to find a starting design, then download both the HTML and the DESIGN.md so the AI gets the recipe plus the finished dish.
20:40 – 23:00
15 · Skills as ingredients (lasers, WebGL, etc.)
Skills are reusable prompt blocks for visual effects — lasers, WebGL/Three.js, skeuomorphism, 3D globes. Meng claims any landing page with a 'laser' skill outperforms because it breaks the generic baseline.
23:00 – 25:20
16 · Live build in Aura + Google Stitch
Meng creates a landing page for a fake startup 'Aura' with the attached DESIGN.md, then tries the same prompt in Google Stitch to compare. DESIGN.md attachment is still new to most platforms — that's the edge.
25:20 – 27:40
17 · Newform — queuing multiple designs at once
Meng demos Newform, his own tool: queue a motion design, a mobile version, a slide deck, all from the same DESIGN.md + skill stack. Comparison to Midjourney's parallel generation flow.
27:40 – 30:00
18 · Skills section deep-dive (63 skills)
Tour of Newform's skill marketplace — skeuomorphic, 3D, batch design, copywriting skills. Each one is just a copyable prompt; you can grab only the colors or only the typography. Vocabulary lesson on font smoothing, body fonts, secondary buttons.
30:00 – 33:20
19 · 1,000 prompts per product
Greg asks how many prompts Meng uses to build one of these. Guess: 25. Reality: at least 1,000, sometimes 10,000. Meng is currently building four products solo (Aura, Newform, DreamCut, rebuilt Design Code).
33:20 – 36:40
20 · Reference → systemize → iterate → remix
Meng's full loop: start with a reference, generate, inspect, systemize via DESIGN.md, iterate (the 1,000-prompt phase), then remix into slides / mobile / promo videos / Hyperframes / Remotion.
36:40 – 40:00
21 · Iteration vs remix
Iteration = small incremental polish when you're happy. Remix = jumping to a new product category. Meng is 90% iteration, 10% remix. 'Soul' is what humans feel across the screen — care has just migrated into the prompting workflow.
40:00 – 43:00
22 · What's changed for designers
Designers do less pixel-pushing and more 'judgment per minute.' Meng cites the Amazon CEO 'one powerful decision per day' line. Designers are now creators — making YouTube videos, marketing, promo material.
43:00 – 46:20
23 · Solo vs team — Meng's holding-company model
Meng builds each product solo for a month (1,000–10,000 iterations) then hands off to his team for templates/videos/expansion. Greg names it: a holding company of products as a one-person business — impossible a few years ago.
46:20 – 49:00
24 · Taste, niche, authenticity (the closing rant)
Meng's parting argument: the only moat is keeping up with what's new. Build taste by surrounding yourself with great design. Niche beats generalist. Anything that looks like another thing loses 10-100x value.
49:00 – 51:02
25 · Second-brain-for-design + outro
Greg's add: most of us have second brains for notes but not for design inspiration — start capturing references like meeting notes. Meng agrees, outro thanks and links in show notes.
Lines worth screenshotting.
- Design.md is a markdown file that stores a full design system — typography, colors, spacing, WebGL — so any AI agent can produce consistent output across every medium.
- The purple gradient problem: what looked stunning five years ago now signals a generic, zero-taste build to anyone who spends time online.
- One-shotting a beautiful design is the easy part; staying consistent across every subsequent page is where 95% of vibe-coded projects fall apart.
- Design drift — starting strong then watching every new section look more generic — is the specific problem that Design.md was built to solve.
- Skills are reusable prompt ingredients, not templates: they give your AI agent a specific capability like lasers, skeuomorphism, or 3D that can attach to any project.
- A designer working with AI today makes more micro-decisions per minute and fewer pixel-moving decisions than at any previous point in history.
- Spending almost $500,000 on tokens building four AI products simultaneously is now a real operating cost for a solo creator.
- Having HTML alongside Design.md matters because the markdown alone can't hold WebGL animation data that makes a landing page actually distinctive.
- Iteration (small refinements to an existing design) and remix (new medium or category) are different creative modes and require different prompting strategies.
- Building a tool for yourself, because no existing tool does exactly what you need, is still the most reliable path to a moat.
- Taste — design, code, or any domain — is developed by surrounding yourself with high-quality examples, not by prompting harder.
- At least 1,000 prompt iterations on a single product is not unusual for someone building at a professional standard with AI.
- Niche is now more valuable than general: a generalized product competing against well-funded companies has almost no shot without a specific angle.
- The only sustainable moat for a small team today is the speed to adopt new tools and models before the large companies absorb them.
One file carries your design system everywhere
What it teaches
Meng To's DESIGN.md workflow: a portable markdown blueprint that keeps one beautiful design consistent across every format and medium, using skills as reusable ingredient blocks.
01Greg's cold open — Greg promises that by the end of the episode 'you will become a designer' — DESIGN.md is the unlock, his friend Meng will demo the workflow.
- Promising a concrete skill outcome in the cold open — 'you will become a designer' — sets a specific expectation that the rest of the episode has to deliver on.
03What is DESIGN.md (Meng's framing) — Meng explains DESIGN.md as a portable blueprint that lets one beautiful design propagate across landing pages, mobile, slides, and motion — solving the 'great first screen, generic everything after' problem.
- DESIGN.md is a portable markdown file that captures a complete design system — typography, color, spacing, components — and travels with every AI prompt so output stays consistent across landing pages, mobile, slides, and motion.
05Five-point agenda + DESIGN.md = soul of design — Meng frames DESIGN.md as 'the soul of a design ported into the agent' — colors, typography, the system — same family as agents.md / skills.md / soul.md, but for visual design.
- Framing DESIGN.md as 'the soul of a design ported into the agent' positions it as a first-class artifact alongside the code — not an optional add-on but the thing that makes every other output coherent.
07One DNA, many mediums — Meng shows a slide and a promo video built from the same DESIGN.md — 'same DNA, different approach' — and points at a community where these markdown files live as free downloads.
- One DESIGN.md can generate a slide deck, a landing page, and a promo video that all read as the same visual identity — the constraint is the file, not the medium.
08Recipe vs dish vs ingredients — Core metaphor: HTML is the finished dish, DESIGN.md is the recipe, skills are the ingredients. Combine all three to get repeatable beautiful output.
- The recipe-dish-ingredients metaphor frames the workflow: HTML is the finished dish, DESIGN.md is the recipe, and skills are the ingredients — combine all three to get repeatable, high-quality output.
09Design Drift — the real problem — Meng names the failure mode: 'design drift.' You one-shot a great first screen, then every later page diverges. DESIGN.md is the cure because the system travels with the prompt.
- The failure mode in AI-assisted design is design drift: a strong first screen that degrades into something generic on every subsequent page because the system never traveled with the prompt.
10Why templates fail vs blueprints win — Templates lock you behind Figma/Framer/Webflow pixels — you can't tell an agent what they mean. DESIGN.md communities share the blueprint instead, so the AI gets a foundational system rather than a literal copy.
- Templates lock you behind tool-specific pixels that an AI cannot interpret; a blueprint shares the underlying system, so the AI gets a foundational design language rather than a literal copy of something it cannot read.
11Cookie-cutter is dead — Greg + Meng riff on Shopify/WordPress/Framer sites all looking identical — like a downtown core with homogeneous buildings. Purple-gradient era is over.
- When every tool generates the same purple-gradient aesthetic, distinctiveness becomes the competitive advantage — and DESIGN.md is the mechanism for injecting distinctiveness at the system level.
12Taste as the real moat — Meng reframes intelligence as 'making ultra-quick decisions about taste.' DESIGN.md is design memory you can transfer between Lovable, Cursor, Codex, Claude Code, Figma — your taste, portable.
- Taste is portable design memory: DESIGN.md captures your aesthetic judgments and transfers them between Lovable, Cursor, Claude Code, and Figma without starting from scratch in each environment.
13Local is the moat — Meng's products generate MD files in nested folders he can open in Codex/OpenClaude — 'because it's local' the agent can read the whole knowledge graph in one shot. Tokens, lock-in, and remote-only tools are the contrast.
- Storing design files locally gives an AI agent access to the entire knowledge graph in one shot — remote-only tools fragment the context; local folders do not.
14Demo setup — Aura, Variant, remix — Meng walks Variant.com and his own Aura — remix-click-remix loops to find a starting design, then download both the HTML and the DESIGN.md so the AI gets the recipe plus the finished dish.
- Downloading both the HTML and the DESIGN.md from a reference design gives the AI the finished dish and the recipe simultaneously — either alone produces weaker output than the two together.
15Skills as ingredients (lasers, WebGL, etc.) — Skills are reusable prompt blocks for visual effects — lasers, WebGL/Three.js, skeuomorphism, 3D globes. Meng claims any landing page with a 'laser' skill outperforms because it breaks the generic baseline.
- Skills are reusable prompt blocks for visual effects — WebGL, lasers, skeuomorphism, 3D globes — and any landing page with even one distinctive skill outperforms the generic baseline because it breaks the pattern.
16Live build in Aura + Google Stitch — Meng creates a landing page for a fake startup 'Aura' with the attached DESIGN.md, then tries the same prompt in Google Stitch to compare. DESIGN.md attachment is still new to most platforms — that's the edge.
- Attaching DESIGN.md to a prompt in any AI design platform is still novel enough that it produces consistently better output than the default — the edge exists because most users do not do it.
17Newform — queuing multiple designs at once — Meng demos Newform, his own tool: queue a motion design, a mobile version, a slide deck, all from the same DESIGN.md + skill stack. Comparison to Midjourney's parallel generation flow.
- Queuing a motion design, a mobile version, and a slide deck simultaneously from the same DESIGN.md compresses what would otherwise be three separate briefs into one parallel operation.
18Skills section deep-dive (63 skills) — Tour of Newform's skill marketplace — skeuomorphic, 3D, batch design, copywriting skills. Each one is just a copyable prompt; you can grab only the colors or only the typography. Vocabulary lesson on font smoothing, body fonts, secondary buttons.
- A skill marketplace where each entry is a copyable prompt — not a locked component — means skills are composable: grab only the color system from one, only the typography from another, and combine them without inheriting the rest.
191,000 prompts per product — Greg asks how many prompts Meng uses to build one of these. Guess: 25. Reality: at least 1,000, sometimes 10,000. Meng is currently building four products solo (Aura, Newform, DreamCut, rebuilt Design Code).
- The gap between 25 prompts and 1,000 is the gap between 'I asked AI to make me something' and 'I built something with AI' — the iteration count is what separates a prototype from a product.
20Reference → systemize → iterate → remix — Meng's full loop: start with a reference, generate, inspect, systemize via DESIGN.md, iterate (the 1,000-prompt phase), then remix into slides / mobile / promo videos / Hyperframes / Remotion.
- The workflow loop is: find a reference, generate, inspect, systemize into DESIGN.md, iterate (often 1,000 or more prompts), then remix into slides, mobile, and promo video from the same source file.
21Iteration vs remix — Iteration = small incremental polish when you're happy. Remix = jumping to a new product category. Meng is 90% iteration, 10% remix. 'Soul' is what humans feel across the screen — care has just migrated into the prompting workflow.
- Iteration and remix are different operations: iteration is small incremental polish on an approved direction; remix is jumping to a new product category from the same design DNA.
22What's changed for designers — Designers do less pixel-pushing and more 'judgment per minute.' Meng cites the Amazon CEO 'one powerful decision per day' line. Designers are now creators — making YouTube videos, marketing, promo material.
- The designer's job has shifted from pixel-pushing to judgment per minute — the same care that used to go into placing elements now goes into prompt decisions, and the output volume is orders of magnitude higher.
23Solo vs team — Meng's holding-company model — Meng builds each product solo for a month (1,000–10,000 iterations) then hands off to his team for templates/videos/expansion. Greg names it: a holding company of products as a one-person business — impossible a few years ago.
- The one-person holding company model — solo building each product for a month with thousands of iterations, then handing off to a small team for templates and expansion — was structurally impossible before AI-assisted design.
24Taste, niche, authenticity (the closing rant) — Meng's parting argument: the only moat is keeping up with what's new. Build taste by surrounding yourself with great design. Niche beats generalist. Anything that looks like another thing loses 10-100x value.
- Anything that looks like another thing loses 10-100x of its value — the moat is not the tools, it is the taste to know when something looks like something else and the willingness to keep iterating until it does not.
25Second-brain-for-design + outro — Greg's add: most of us have second brains for notes but not for design inspiration — start capturing references like meeting notes. Meng agrees, outro thanks and links in show notes.
- Most people have a system for capturing ideas but not for capturing design references — treating visual inspiration with the same discipline as note-taking is the starting point for building taste deliberately.
Terms worth knowing.
- DESIGN.md
- An open-source markdown file format, popularized by Google, that encodes a complete design system — including typography, color palettes, spacing rules, and component patterns — so it can be attached to AI prompts to produce visually consistent output across web, mobile, slides, and motion formats.
- design system
- A documented set of reusable design rules, components, and visual standards — covering typography, color, spacing, and UI patterns — that ensures visual consistency across all surfaces and formats of a product.
- vibe coding
- A colloquial term for AI-assisted development where non-engineers build apps and websites by describing what they want in natural language — prioritizing speed and output over traditional engineering workflows, but often producing generic or inconsistent visual results without design guidance.
- MRR (monthly recurring revenue)
- A standardized metric for subscription or recurring businesses representing the total predictable revenue generated in a single month — used to measure growth and business health independent of one-time sales.
- WebGL
- A JavaScript API that enables interactive 3D and 2D graphics directly in the browser without plugins — used in design systems to specify rich visual effects, animations, and data visualizations on the web.
- skill (design)
- In this context, a reusable, shareable prompt or instruction file that encodes a specific design technique, component, or visual style — part of an emerging marketplace model where designers share repeatable AI-ready methods rather than static templates.
- design token
- A named variable that stores a single design decision — such as a specific color value, font size, or spacing unit — enabling design choices to be defined once and applied consistently across all platforms and outputs.
- motion design
- The discipline of applying animation and movement to visual design elements — transitions, micro-interactions, scroll effects — to make digital products feel more dynamic and polished than static layouts.
Things they pointed at.
26:16productDreamCut (Meng's upcoming Mac app)
03:50toolReply Slides
03:54toolHyperframes
16:22toolCodex (OpenAI)
03:31toolOpenClaude / opencode
03:04toolGPT image (new)
41:17bookJeff Bezos 'one powerful decision per day' (paraphrased)
44:20bookSteve Jobs 'go fast alone, go far together' (fake quote, attributed)
Lines you could clip.
00:30
“By the end of this episode, you will become a designer. You're gonna be able to create better design than 99.99% of this planet.”
09:24
“The HTML is more like the finished dish, and the MD file is more like the recipe. The skills are like the ingredients.”
10:57
“How do you solve this problem, which is design drift?”
27:00
“It's hard to tell if this is made by a human or by AI, but I can guarantee you 100% that this was made by AI.”
33:09
“Can you guess how many prompts I use to build these products? At least a thousand.”
39:35
“AI is not making me lazier. AI is making me more work more.”
40:59
“We're doing a lot less of the moving pixels, moving rectangle and resizing things. We're doing a lot more judgment per minute.”
48:49
“Nowadays I see a website with purple gradient, I don't wanna even wanna scroll anymore.”
47:38
“Niche is the keyword here because you cannot compete anymore with people who are generalized.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogystory
00:00There's a way to create beautiful designs, beautiful websites, beautiful apps, beautiful motion, and I just learned about it. And it's through this thing called Google MD.
00:10It's using skills, and the entire process of this is what we explain on this podcast. I had my friend Meng To, who's one of the best designers I know, come on and share his entire workflow to how to create jaw dropping beautiful designs that by the end of this episode, you will become a designer.
00:31You're gonna be able to create better design than 99.99% of this planet. And what does that mean?
00:37That means that your startup, your idea, your app is gonna have a higher chance of getting people to install it, to buy it, to share it, and we all can use that. So enjoy the episode, and I'll see you in there.
00:59Meng, welcome back. Second time on the pod. Uh, by the end of this episode, uh, what are people gonna get out of this?
01:07Google just came out with Design. Md, and this is one way to create a blueprint to take one beautiful design, whether you made it or not, and then spread throughout the all the formats and the mediums for your startup idea.
01:25Right? So you can turn a beautiful design system that is created by a master in their craft, like a designer, and then you can turn that into motion design, a landing page, different sections of the landing page, uh, mobile design, and so on.
01:43So a lot of people struggle right now with, you know, this idea, you know, they start really strong with one shot and then suddenly they get into the the other pages and they suddenly have something more and more generic.
01:59So this is what we're gonna be learning. I'm gonna teach you all the tricks, remix, iterate, what is the design system, typography, colors, all the things that people struggle in design.
02:10Perfect. In in your world, feel like a lot of people are talking about DesignMD. Uh, in my world, not that many people are talking about it, but they want to be able to create jaw dropping beautiful designs even if they're not a designer because they're, you know, we're using agents now, we're building startups, we're taking ideas out of our head, but you know, we don't want a purple vibe coated website or app.
02:36We want something that's beautiful, that's consistent, and so I'm excited for you to explain what DesignMD is.
02:44I'm excited for you to explain how you can actually get inspired by other designers and use some of their systems to create something of your own. Meng, thank you so much for being generous with your sauce and and tactics and tips and tricks.
02:59Uh, there's no one I look up to more than you in this in this regard, so, uh, thank you.
03:06Yeah. I I just wanna mention something before we start. People don't realize how hard it is to to get the right distribution, the marketing.
03:14And I wanna thank you, Greg, because, you know, remember the first time I came to your podcast, I started with a $3,000 MRR, and it shot up to 15,000 MRR just by going through your podcast.
03:25And, obviously, there's a ton of value over time. So, you know, to those who don't know, like, going to, you know, Greg's podcast is like the bucket list, like, the number one thing you wanna, uh, you know, reach towards to.
03:40And, yeah, I prepared presentation, you know, slides and all that stuff because, like, you know, Reply just introduced slides and, you know, we started doing, uh, you know, hyperframes. We started doing, uh, remotion.
03:53So I prepared all of this. Uh, as a designer, I feel very strongly about this.
03:58So let me just share my screen. First things first, I realize a lot of people are vibe coding their apps, they're vibe coding their startup marketing, and so on.
04:09Everything I do today, including this tool that I'm showing you, which is my own Notion with slides, is VibeCoded, and I made it by myself.
04:21So I also wanna say that all of these images are all generated using the context of the title, and I'm using the new GBT, uh, image too.
04:36So with that said, you know, I prepared the slides, and I wanna go through one by one what is design.mt.
04:46So but first of all, you know, I just wanna also preface on what we're gonna be discussing, you know, the things that I wanna teach everyone about design and how we get there, and then we're gonna end with a demo.
05:00So these are the kind of five main points I wanna I wanna go through. And then we we're gonna get started.
05:08Uh, design, as I mentioned before, design.md is just open source, and I can find the tweet right here.
05:16They have a ten minute video. You can watch it. Essentially, it means is that just the same way a lot of people are using agents.md or skill.md or soul.md for those who are using Cloudbot or in in this case, it's it's OpenCloud now.
05:38So, basically, this is for designers. And if you wanna take and you wanna bring that to the agent and you wanna bring a design system, the colors, the typography that makes a design beautiful, well, you port it and you put it all into an MD file, and then you put it into your prom as an attachment, and then boom, you have a beautiful design.
06:03So this is what design.md, and you will see extremely efficient ways to use this across multiple medium, not just web design, but also, you know, like, uh, you know, like motion design nowadays and slides.
06:24So,
06:25you know, for exam Quick break in the pod to let you know about a free workshop I'm doing that answers some of your biggest questions. The first is, in the AI age, what are some categories that are ripe for startup ideas?
06:38I'm gonna outline a bunch of those different categories, the ones that I think the most opportunity is. Second is there's a million AI tools out there, which are the ones that matter when I'm trying to build a business? I'll walk you through that.
06:52And lastly, how do you actually build a business with some of these AI tools? So I'm actually gonna show you live how I do this with my ideabrowser.comteam and I can't wait to see you there.
07:03It's this Thursday at 12PM eastern and if you go to the show notes, if you go to the description, you can click a link, RSVP, and learn a few things.
07:13I hope it'll get your creative juices flowing, and I'll see you back at the pod. I'm gonna give you just some quick examples of what I created. So for example,
07:22when you have a design, right, you can sort of, like, take one design and using whatever medium, like, whether it's React or HTML, you can create, like, a a nice promo video or you can create, like, a slide.
07:40And this is all using the same DNA, which is the design. Md.
07:46So as you do this, for example, this is a slide and this is a promo video. You can see that this is two different approach to a design. And this is the result, and, uh, if you want, there's a ton of resources that allows you to save those design dot m d.
08:04So one of them, obviously, I created them myself, but they're totally free. You can just download them. Uh, right now, my Internet is super slow.
08:14So here, you can see we have a lot of these design systems, and it's essentially just one file.
08:22You click here, you download it, and it it contains all of this visual stuff, typography, colors, spacing.
08:32And if you look at the, you know, the the the markdown itself, if you've used markdown, it's essentially a structured text with tables, uh, titles, and code.
08:46And you can if you read through all of this, you have the weapons and the arsenal to essentially prompt efficiently.
08:55And you can sort of commit this to memory, and then it becomes your design.
09:00So it's very simple. You add all of this to prompt.
09:04Right? And then it becomes the the HTML or the design.md.
09:11The the sort of a finished dish, and the m d file is more like .
09:20The skills are like . So you put it all together, and it becomes your design, which is what I just showed earlier.
09:32And that's a big deal because, you know, why like, honestly, I want to have you on the pod is, you know, it's such an advantage to to having scroll stopping design.
09:45Like, the video you shared, like the the the m p four there that you shared, like, that was that was beautiful.
09:57And so I guess my question is, like, how do you if you're not a designer, coming up with these systems yourself, like, from scratch is really hard.
10:10So how do you what's the process for for for finding, you know, for finding brands and identities that resonate with you?
10:24Right. So, you know, coming back to the slide, right, oftentimes, you know, people, they one shot something.
10:32They saw, like, a a beautiful prompt. Oftentimes, it's bigger than the code. They copy and paste that.
10:37They have no idea how it's done, and they also have no leverage on how to change it. So the first screen is not the hard part. So, you know, right now, it is you kinda feel like you're the one who's always, like, telling AI what to do, and you don't have control.
10:54And the question is, like, how do you solve this problem, which is a ? Okay? You you you start something, you're kind of satisfied with it, but then you get into the other stuff, and it becomes completely different.
11:08So that's that's why we need design.md. And, you know, you don't want to end up something like this.
11:15You want to to end up with a system, a process where you can bring whether it's a screenshot, an HTML, a design.
11:25Md, or all of those ingredients and recipes.
11:30And back to your question, Greg, there's a ton of resources that share all of these things, but designermd is new. Most places, they share templates.
11:41Right? We all know, you know, like, you go to a template, and obviously, I'm taking my mine as an example, but you can go to any, uh, of the templates that you find on v zero, Lovable, um, you know, all of these other place Framer, Figma community for for people who are more familiar with designs.
12:03And the problem with those is that they are highly close behind the techniques and the the rectangles and the pixels of Figma or Framer of Webflow.
12:15But, you know, then how do you tell that to an agent? And that's the big question. So nowadays, you can go to new communities that share the actual blueprint, which is the design.md.
12:31So for example, I want a design that looks like this, that has this animation, that has the blue color, that has these systems of, uh, you know, beautiful typography and selections and buttons that, you know, I don't want everything.
12:48I don't want the whole template. I just want something to start with that I can tell my AI to start with that and then be consistent across the board.
12:58Now if you only provide a portion of it or something too specific, the AI tends to sort of, like, take that too literally.
13:09But if you provide something more, uh, like a foundational system, like a process, agents.md and what and whatnot, then you get something that gives more flexibility to your system when you're gonna be working on new designs or new mediums or, you know, you wanna be consistent across the board.
13:31Cool. Yeah. I mean, when I see this, like, first of all, this is gorgeous.
13:37But what I would wanna refrain from is, like, to your point, duplicating that and just creating the same cookie cutter website.
13:49You know? Because there's I'm sure you see this all the time. Like, you you know, you come across a website.
13:54It's a Shopify. It's a WordPress. It's a Framer.
13:56It's whatever. And you're like, I've seen this website before. Right?
14:00I've seen this website, but it's another brand. Right?
14:04And that like, cookie cutter doesn't work anymore.
14:09And it's the same reason, like, you know, you go into a a downtown core city and, like, all the buildings look the same. Right? And and you feel like there's, like, homogeneous cityscapes that that are existing now.
14:25Yeah. And I fully agree with you. Everything is so high.
14:29Like, the baseline is so high nowadays, but also the baseline is also very generic. What used to be wow five, ten years ago now is generic, which is funny because, like, if you look at the purple gradient, everyone was like, wow.
14:43Purple gradient. I wanna use that. But nowadays, if you see something with, you know, purple gradients, you just run.
14:50Uh, so I guess what I'm trying to say here is that we have evolved so much, and humans have a way to solve new problems and to adapt.
15:00And I think that's the real intelligence. And so I think the question nowadays for me is how do you deal with taste?
15:09Right? How do you design something?
15:12How do you make ultra quick decisions? Right? To, uh, take something, whether it's the ingredients, the recipe, the secret sauce.
15:22For me, the sweet secret sauce is is the design. Right? Because, you know, you don't want people to run away from your landing page.
15:31You don't want them to, uh, to to to look at your brand and like, oh, shit. This feels like, you know, just like the 1,000 start up that I've seen over the past week.
15:44So if you if you focus on design enough, right, if you find these systems that you can, uh, bring to your own system that is flexible enough, then you can create something way more powerful.
16:03So I like to think of Designer. Md like a design memory. And this is great because it it it can transfer between platform.
16:14So if you're working in Lovable and you like what you've done or in cloud design and then you wanna move it to Figma or you wanna move it to, uh, you know, to Cursor or to you know, nowadays, I use codecs a lot.
16:32I know a lot of people use Cloud Code, but you can totally do all of that stuff because you can carry this this Designer. Md, and then you can commit it to memory by telling your AI, your agent, to remember.
16:45That I mean, I do that all the time. Remember this. Remember that.
16:49And I think agents nowadays are are doing a pretty good job at, um, you know, trying to remember the workflow that you've just done.
16:59Right? Because that workflow is unique to you. Like, a lot of things that I make nowadays, I'm I'm sure it applies to everyone out there who is making their own apps.
17:09First of all, for me, when I make apps, I look at what is the moat.
17:16What what is something that I can do with my app that Notion cannot do or Figma cannot do? For me, that mode right now is the fact that everything is local.
17:27So for example, this is all local. And because it's local, it it it generates these MD files across folders and nested folders that I can just open in codecs and open that folder and and tell my my agent, look.
17:45I have a Greg Greg Eisenberg podcast. I want you to prepare all of these screens.
17:51I want you to prepare, you know, 10 sections or 10 notes or do research, and it has all of that knowledge, and it can create all of those files in one single go, which I think is such a big moat nowadays.
18:06Right? You can't like sure, some tools can do that, but one, you have to deal with tokens, you have to pay for them, you cannot run locally, and you don't have your own workflow embedded into it.
18:18But nowadays, you have something like OpenClaw or or Codex or CloudCode that can do all of that, and it's so scary how good it is.
18:27A 100%. So what you know, you mentioned, you know, Cloud Design.
18:34You mentioned, you know, all these different platforms. You know, we're talking about Google, so Google Stitch is another one.
18:42Like, can you show us how we can how we can go from, you know, MD file plus HTML to creating something?
18:55Sure. First of all, I just wanna start with sort of, like, you know, getting the MD file, but also focusing on the design aspect.
19:07For me, it's really important, but not everyone's gonna do that. But I think it makes sense to start with that.
19:13So for example, you know, I don't know if you know varian. Varian is also very good.
19:18Yes. Varian.com. You know, I like tools like this because they allow me to get into the sort of creative stage, which I think is incredibly important.
19:29And essentially, what you wanna do is just to remix, you know, click one button, remix, click one button, remix. You find something that you like, you remix, and so on and so forth. So that's why, you know, for me, as someone who build aura, I also find it incredibly enticing.
19:43So I go to community, right, and I find something that I like and, uh, you know, I just I just remix it, uh, really quickly.
19:52In one click, I I have a bunch of skills. Like I mentioned before, skills are like ingredients. And, you know, if you go to to to some of these skills, you're gonna see that some of them are so so powerful.
20:05And this is your moat in design because you don't wanna look generic. You don't wanna have purple gradients. So the only way to do that is, for example, you you create a skill for lasers.
20:16Okay? So I know it sounds funny, but whenever I create a landing page with lasers, everyone clicks on it. I don't know why, but people love it so much because they love special effects.
20:27Right? If you go to a movie, what you know, you wanna see special effects. You wanna see, you know, the Avengers kind of, you know, doing these grand things.
20:36So that's the mode. Right?
20:39Like, nowadays, just having the typography, just having the colors is not enough unless it's a secondary page or unless it's something that is very serious.
20:48But if you're talking about a landing page or a promo video or, you know, a slide, you need a moat. So, uh, yeah.
20:56You know, like, remix super fast, blah blah, you know, like, uh, you're gonna remix this, click one button, you know, and then do that until you're happy. Once you're done with that with that flow, then you get into the prompt.
21:12Right? So each skill has a prompt, so you can copy this. And anything all of that stuff, like DesignMD and skills are free.
21:20And, uh, this is not why you should be paying money. But you you pay money because of tokens.
21:26You pay money for the finished results such as a template, um, and sort of like the whole sort of automation and whatnot.
21:35So, you know, you're gonna go to a template, you know, like, uh, I don't know.
21:41Let let's go to to this one. And and then you're gonna download both the HTML and the DesignerMD.
21:51The reason why I'm saying HTML is that's, you know, design.md may not hold all the information that that you need to keep to create your first result.
22:01It does hold sort of, like, the typography, colors, the foundation, the spacing.
22:08Some of them, including my own, holds also the WebGL, which I find for those who don't know, WebGL is the animation that powers the laser. Right?
22:17Without WebGL or Three. Js, which is for three d, you don't have any of that stuff. You only have the typography and some of the reveal animations and some of the rules.
22:27So that's why these two elements here for me is really important, and that's what gives you these beautiful animation that makes your site a lot more unique. It goes from, you know, zero to 50 or 50 to 80 real fast.
22:43And so, yeah, I would download the two of them. And, you know, back to here, so we already had the demo, so I might as well show you guys. Uh, on Aura, you can go to design.md.
22:56This is why where you're gonna find all of this stuff, the design, and then you you take something that you like and you add that to your prompt.
23:05So the same way that I download design.md in HTML, I add it to my prompt, and now I'm ready to to tell what I want for my landing page.
23:16So for example, I can say something like, okay, create a landing page for my startup call, you know, I don't know, like, Aura.
23:28And, uh, I know, really original. So and then it's going to be a chat app that is, um, you know, that that is using AI and that ships stuff to people, uh, to their email, whatever.
23:46So let's see what it does. Okay. So now we have the design.md.
23:52We have the HTML, and then we we're just gonna create it. Let's hope for the best. But, uh, you know, it's got this this obviously, you don't have to use Aura.
24:02You can just use whatever tool that you want. I just wanted to show you that that's essentially what you're supposed to do. You download the two files or just a design.
24:11M d, however you want that to be, and then you start creating your, uh, your page, uh, through that system. And from there, you can generate more, you know, elements, more sections, more because you have the designer.md always with you.
24:30And also, you can bring that to another platform. It's not specific to one platform.
24:36And what's your take on Google's Google Stitch? I haven't I haven't gone deep in it.
24:42Right. So for me, honestly, as someone who builds a startup using tokens, you know, it's it's really unfair, to be honest, because they they're they're able to give all of these stuff for free.
24:58And I just realized, you you know what? Like, I spent almost half $1,000,000 in token for my startup right now. Right?
25:06It's it's insane. So, yeah, you can do basically the same thing. Uh, I don't know if this is the correct one, but, you know, like, uh, create the landing page for Aura.
25:21And, uh, you know, you're gonna select web, and you're gonna select 3.1.
25:28Now what makes it different is that it has an infinite canvas.
25:34Oh, I don't know what's going on here, but it's it's also doing stuff differently.
25:41And this this concept of, like, attaching DesignIMD, which I feel very strongly about, is still, like, new to a lot of platforms.
25:49So as someone who works in AI, your moat is basically you have to be new. You know? You have like, I I don't I'm sure any anyone can relate, but you always when there's a new model that comes, you have to talk about it.
26:02Otherwise, you're kind of losing a lot of that sort of edge that you have all over all of these companies. So the same way with tools, you you know, and and the same way when you build a start up, you have to utilize all the best technologies and tools.
26:22And in this case, we're using a Designer. Md. It's recreating exactly that foundation that we saw on Aura.
26:30And, you know, it's going to create the same consistent design whether it's from here or on Aura.
26:39Uh, in this case, I used two different designs, so that's why we have two different designs. But if, you know, in this case, I'm using Auralist and the other one, I use the one that I I I click earlier. But, you know, here, I can just say, you know, create the footer, and it's gonna create, like, add the footer, and it's just gonna create the footer.
26:57But you can see the design using this technique gives me exactly what I want. It also, uh, you know, it also adapted to my to my design, to my brand, to my copywriting.
27:09And then from here, you can use skills. Right? Like, for example, I I love to use skills like if you go to skills, copywriting skill.
27:20So this is going to improve your copy by a 100%.
27:26You know, it's gonna look at your text, and it's going to to make it so much better. So skills, Designer.
27:35Md, HTML, all of these things, if if you merge all of them and blend all of them together, you create something really, really powerful.
27:46And, you know, to visualize all of this, you know, this is a new tool that I built, which is called Newform. You know, you can sort of, like, in one click, I turn this, you know, which is a motion design.
28:00Right? I go to prompt, and I have all of these skills that I already saved.
28:05And then I can just click, oh, I want laser. I'm gonna click on laser. And it you know, I can just create another one and another one.
28:12So I love the the idea of queuing, which is why we have a different products is because, like, queuing is is such a powerful idea that I use all the time. Nowadays, I create, like, three, four products. At the same time, I have, like, multiple chats with sub agents on on across all of them.
28:29So that's why, you know, I I wanted to create something way more focused on the creative aspect. And on top of that, we also have the remix type. Again, all of these are prompts, by the way.
28:39Like, I'm I'm not, like, you know, keeping anything behind closed doors. You don't have to use my tools at all. But, you know, I just wanted to show you visually how how it is like to turn something into mobile.
28:52Right? Or, you know, if I go here to mobile, you're gonna see that I have turned my designs into mobile design.
29:01Beautiful. You know, I mean, if you look at this, it's hard to tell if this is made by a human or by AI, but I can guarantee you a 100% that this was made by AI.
29:13And from here, I can just say, okay. I want a slide deck. Okay.
29:17I want, you know, a motion design. I want the hero section of a know?
29:24And then I'm doing six design at the same time, which is so good so good for the creativity.
29:29It kinda reminds me of how mid journey works. You know, on mid journey, you kind of I haven't used mid journey in, a year and a half maybe, but, like, that that concept around you can, like, generate multiple things.
29:44And what's really cool about this is, like, you get into this flow state of, like, okay. I want this. I want that.
29:49I'm gonna queue this up, and it's really fun.
29:54I I agree. Like, the concept of queuing is not new per se. I'm I'm just saying, you know, a lot of the things that we do see here, it just transformed into mobile.
30:05And a lot of the things we do can just be transferred so easily because we have this this concept of, like, skills and design. Md.
30:14Right? Yeah. So, you know, uh, one of your guests, for example, was talking and he's he he calls himself, like, a a skills maxi, which I find it really funny because I totally agree.
30:24Like, for me, I don't put anything in my agents.md because I have my design.md, which I use per project, and then I have my skills, which I use per workflow.
30:36And if you only change the agent.md, oftentimes, you know first of all it costs a lot more tokens but also it's too general like it doesn't apply to every single workflow out there so here I added the laser and here I turn it into a slide you know like again it's so so fun to queue and to sort of, like, expand your creativity across all of this stuff.
31:02So you have something like Aura Lovable VZero, which allows you to chat and sort of, like, evolve into something into a product. But I think what people sort of, uh, underestimate a lot is the design aspect which is, you know, how do you come up with new sections that are unique.
31:19Right? How do you, uh, you know, work on the creativity and how do you learn about all of this stuff?
31:27Because if you don't learn this stuff, you know, you're you're just focusing so much on the chat and you're just chat chatting chatting and then giving commands in the AI. The AI do do all the work for you, then you're not making any decision or at least you're not making the best decision for your product.
31:44Can you quickly go back? I'm just curious more than anything, like, the skills section of Newform, I think it's called. Yes.
31:52Like, it's what happens when I click there?
31:56Yes. So you have access to right now, have 63 skills, but I'm adding more every day.
32:01So for example, you have a skill for skeuomorphic design. For those who don't know, skeuomorphic design is kind of like the the realistic design that Apple used to use before the flat design. And it's a prompt.
32:13It's just a prompt and you can copy this prompt and you can just tell your agent to use, uh, you know, to use this skill and then to apply that.
32:24So for example, you're gonna have something a little bit more embossing, know, something like like this design for example.
32:31You know, it has some of these, you know, shading and stuff like that. A lot of people still love that.
32:38I love that when it makes sense. So for example, if you do like a DJ app, you might want something with a little bit more realistic. You want a button that is lickable as Steve Jobs says, well, yeah, you need a skeuomorphic design.
32:50Right? And we also have a skill for, uh, you know, for the three d. So, you know, this this little globe here that you put, well, that's three d and then you can move that on top of the text which makes it look really nice.
33:04So a lot of like, every skill can be anything, you know, it can be like a batch design, uh, like this one, and you can have, uh, you know, all of these are also copyable.
33:18So for example, if you only want just the colors, can just copy the colors or just want the typography. So not only do you learn how to sort of like add to your arsenal all of these keywords that you did what is font smoothing, what is sound font, what is system font, what is a body font, or what is the secondary color, what is a secondary button.
33:41So all of these things are really useful when you, you know, chat with your AI agents and, you know, sometimes I condense it into just one word.
33:52Okay. Fix the spacing. Fix the gap.
33:55You know? Fix this. Fix that.
33:56And it's just so powerful to learn all of this stuff because it's all part of the Designer. MD and it it holds so much richness to how you can manage your design.
34:07And it must be cool for you because you've been, you know, you've been teaching designers for years now and these are, you know, mostly, I mean, trained designers.
34:18Now anyone is a designer.
34:22That's such a good point, honestly. You know, when when I first started, nobody was using like, most designers were not coding.
34:32And nowadays, everyone is coding including nondesigners. Uh, you know, before, like, ten, twenty years ago, everyone was using Photoshop and then Sketch and then Figma.
34:43But nowadays, everyone is a designer, and everyone is kind of, like, downloading a template or an HTML or a design. M d, and then suddenly they have the most beautiful landing page. Sure.
34:55They one shot it. Sure. They don't know how to change it or make it better, but they still have an amazing one shot, you know, one shot prompt.
35:06Uh, but what I wanna teach here is the whole workflow so that you don't end up with just a one shot shot thing. And how do you you know, can you guess how many prompts that I use to build these products? Just give me a number.
35:18How many prompts do you think I use? It's definitely not one shot.
35:23I don't know. A couple dozen? Twenty?
35:26Twenty five?
35:27At least a thousand. Yes. This is how obsessive one can be Wow.
35:34When it comes to creating. Yes. Like, I'm building four products right now, you know, Aura, Newform.
35:43Well, this I don't really count it as a product. I have DreamCut, which I'm gonna be real revealing soon.
35:50It's a Mac app. And, you know, I'm also rebuilding design code from scratch.
35:57So four product all at once, and I'm not even including a side product like this one, which, honestly, I love using it just because I'm able to tell my agent to write everything for me, and I'm able to just generate these images, these YouTube covers using a history of my portrait that I save into this because there's no tool that does this.
36:18Right? Like, you you build tools literally because nothing else is doing it.
36:23And there's nothing else that has a slider that says, okay. I want this to be exactly or I want this to be loosely inspired.
36:30Like, have you seen this UI before? I've never seen that. And I see a lot of products.
36:35Exactly.
36:36Yeah. And the the only reason why you we're able to kind of get there is because we're building something for ourselves. Right?
36:44And it's such a powerful thing. So if you look at the whole workflow, right, you you start with a reference.
36:51Every every designer, every someone who starts drawing something, they always copy first. They always start with reference to design the MD or in in most cases, they're just taking, like, copying a prompt some somewhere.
37:02And then we generate, we inspect, and then eventually using the designer MDU systemize. And then like I said, you iterate.
37:10This is where I get, like, to a thousand iterations. And then you, you know, and then you remix it to turn it into more mediums such as the slide, you need marketing material, Instagram slides, promo videos, you need like intros, you need, you know, something for your hyperframes or for your remotion.
37:29Whatever it is, you're essentially remixing at this point because you're happy. Because you you've done a thousand iterations, you're happy with your product, you're happy where it's going, you you feel there's a salt to it.
37:41Because I think most people, what they tend to forget is that humans, they don't see the soul across the screen. We do. Okay.
37:49We like, we're innate to that. When you you you can feel I don't know how how to say it, but Steve Jobs said it very well. When you use a product and when you really, like, take a magnifier and you look into it, you really feel the care that is put into it.
38:07It's not because we have AI that we we can just skip all of that care. Right?
38:12It just happens that that care is now residing into the prompting and the sort of the workflow, the club, the, you know, the open claw, you know, the things, the thousand iterations that I do.
38:25I've never worked more more in my entire life on more products on on, like, on more problems in my entire life than now. AI is not making me lazier.
38:38AI is making me more work more. And, yeah, you know, and, you know, we we get into the expansion and we export our product and it becomes like a final product that we can ship.
38:51So iteration versus remix.
38:54This is a very important concept. Like I like I mentioned in my slider, iteration is when you're happy with the result and you wanna kind of like boost small incrementations.
39:03And remix is when you wanna, you know, build a new product category, for example. So I love the idea of iteration more, and I use this, I would say, 90% of the time and 10% of the time. And then, yeah, I already did the did the demo.
39:18Okay. So we can talk about what designer what what has changed for designers. So what what I believe is that everything is changing, and we're doing a lot less of the moving pixels, moving rectangle and resizing things.
39:37We're doing a lot more judgment per minute. In fact, you know, for me, the agent is more like someone who's, uh, doing a lot of the sort of the moving pixels and the coding, and then I have to make the right decisions.
39:53I remember the CEO of Amazon was saying like if you make a really good, like one powerful decision in a day or a few powerful, then that changes your entire business. And I think, you know, I agree with that. You have to make these extremely powerful decisions and also you have to make these incredibly small decisions as well just because the AI is not intelligent enough to make the smaller decisions yet.
40:17But eventually it's gonna get bigger and bigger. The decision is gonna become bigger and bigger. Um, and then, you know, that's kind of what's changed.
40:25We're we're shifting our, uh, you know, our creativity and we're shifting to something that is more, you know, uh, looking for a high level.
40:35I've never done more marketing in my life. I've never done more YouTube videos in my life or or slides and and promo videos. So yeah, this is what's changed.
40:45We're we're now everyone's like a creator, everyone's like an Instagram, you know, a filmmaker of some sort.
40:51So this is, you know, I think this is what's what's happening. There's so many ideas that we can create.
41:00And, yeah, you know, don't wanna get too too deep into the slides. I wanna get to your questions, obviously.
41:06Yeah. I think, you know, just your point around, like, you've never done more in your life.
41:11I think what's also interesting is, like, you're you're a team of one. Am I is that correct?
41:17Like,
41:18is that am I correct? I'm I'm a team of many, but not because of what I'm doing right now. Right now, I'm kind of like a team of one.
41:26Like, all of these products I'm showing right now is me building them by myself. Yeah. But the reason why I still have a team is because I started in the days of Design Code and I built this team that I grew really fond of.
41:40They started from nothing. They haven't gone to school like myself. Haven't gone to school to study design.
41:46I drop out of college and then I learned everything on the field. I started my first job at 18 and then you know started my first company at 22.
41:56So you know a lot of like everyone in my team started from nothing And, uh, you know, I feel very attached to them, which is why I have them with me.
42:08But you're right. You know? Like, all like, all of these products, I started by myself.
42:12I work on them for a month, a thousand iterations, maybe more to to some some of them 10,000 iterations and, uh, I arrive to a point where it's mature enough and then we get to the team to to build templates and videos and and and sort of like the material that expands on that.
42:30And, you know, I came across this this quote, this fake quote from Steve Jobs on his on his final, you know, bed, you know, when when he, uh, he he was passing.
42:44It's it's apparently, it's a fake one, but it was saying, like, if you wanna go if you wanna go really fast, you start alone. You start solo. But if you wanna go far, you need to be in a team.
42:55So that's what I think right now. Like, most of us, we start solo because we cannot go any faster.
43:02Right? We we're, like, operating a fleet of, you know, ten, twenty agents all at once, and that's great.
43:10But eventually, you need to get people with you. You you know, you need to be part of a mission, and then you need to sort of sort of like, you know, make these videos that feel more authentic because people crave authenticity.
43:26They want real people. They want real craftsmanship behind all of these pixel art at some point because, you know, the cracks are gonna show if you only use AI the, you know, the whole way.
43:38Yeah. I think there's I think there's two types of people. I think there's the first type of person, which is what you're saying, which is like, go build product market fit, start scaling, and then from there hire a team.
43:50And the second type of person is the person who just loves being a lone wolf, you know, who just loves doing everything themselves. And I think what's cool about what you're doing, you know, you're essentially building a holding company of products as a one person business.
44:10Yes, you you know, some some of your businesses you're scaling with, you know, with people and stuff like that, but you know, the majority of it is is being run by one person, and you couldn't do that a few years ago. And I think more importantly than, you know, within the context of this episode, it's like you're creating beautiful products.
44:28Like, these products that you're creating are stunning. They're nuanced. They look really good.
44:33And I wanna end just with, you know, what what you know, in terms of building beautiful products, you know, we talked about DesignMD.
44:42We talked about skills. We talked about a lot of these other tools, Variant, Aura, some of the new tools you're working on.
44:50Is there anything else you wanna leave people with in terms of how you can create jaw dropping designs?
44:57Yeah. So like I said, your only moat is you being able to catch up with what is going on.
45:04Because if you look at everything that is legacy, that is already solved with all of these big companies that have millions and and nowadays hundreds of millions in the bank to tackle on these big things.
45:16So as people who are starting new ideas, the only mode that we have right now is just we can, you know, just click a button and we can say, we'll you know, we found this new model.
45:28It literally, uh, cooked all the designers and whatnot. Uh, it's funny.
45:33I know. I know we all hate it and it it becomes a meme, but at the same time, it it reflects the reality in which we are today, which is that that's how we survive as a space sheet.
45:45Right? We cannot compete anymore with the big companies and we have so many ideas in our head and we can finally realize them with just a few prompts at at least to get to an MVP.
45:58And that is usually, uh, good enough to launch a product. And there's just so many ideas because of it.
46:04So for example, Design. Md just came out and I already built like a bunch of features. And I could totally build a startup out of that if I wanted to.
46:13Right? Uh, QA, right?
46:15Uh, design, uh, memory, uh, this is what's solved a lot by by OpenClaw nowadays is when when the idea is like it has the whole memory inside your disk, inside your computer. You know, reusable taste marketplace.
46:32And I wanna leave with one final thing which is that taste is the real value here.
46:41And when I say taste, I don't necessarily just mean design taste. Obviously, design taste is very is very cool and is very powerful and I think that's my secret sauce. But for a lot of people, your taste can be, you know, like if you look at Barista, the taste of going of making so many cup cup of copies every morning, you eventually knows exactly how to do it.
47:02Taste in coding, taste in building apps. You have to just be surrounded by good design.
47:10Right? You have to look at good designs, not just be served by it, but also to look for it everywhere. You have to use every app in your niche.
47:21You have to follow every, uh, you know, makers and creators in your niche. And niche is the keyword here because you cannot compete anymore with people who are generalized.
47:32It's too difficult. Nowadays, we want authenticity, we want niche, we want taste, and if something looks like another thing, it it kind of reduced the value by like almost 10 to a 100 x in my opinion.
47:46Right? Like nowadays I see a website with purple gradient I'm like I don't wanna I don't wanna even wanna scroll anymore and I'm sure a lot of people feel the same way when we see like a website that looks just like another one. So yeah develop your taste, rinse and repeat, you know, follow what's going on and you know, get get that to that moat that you have and learn the workflow.
48:10Like everything that I taught here, get your designer MD, get the the templates, learn about the skills and the prompting behind the skills, and then I think you're gonna be doing just fine.
48:22Mhmm. I love that. I also think
48:25a lot of us have second brains for our notes, our meeting notes, ideas that we have, but a lot of us don't have a second brain for design inspiration. So I think the idea of when you see something in the real world or on the digital world where you're like, this is really cool, capturing that and putting it somewhere where you're like, you can you know, when you're when you're trying to create something new, you can refer back to your second brain for creative inspiration.
48:58What do you think about that?
49:00A 100%. I mean, I think whether you call it a second brain or your creative side of the brain or, you know, something that keeps you going because, you know, we crave purpose.
49:12We crave to be praised. And you cannot get if you you cannot get there, uh, you know, you can get not get to that purpose if you don't somehow get excited by it and and kind of explore something new and adapt to a new reality.
49:27And I I think, like I said, taste is so so important. This is where we develop that we make all of these micro decisions super quickly. Uh, in in an instant, you can tell right away if the design is good or not or if there's a care behind it or not.
49:43You know, you touch like a phone, like an iPhone, and you're like, you feel right away this is extremely high quality and you wanna use this, uh, as your, you know, as your main daily driver. So, yeah, you know, please use your your your creative brain.
49:57Please use your second brain. Please use your, uh, you know, your taste every step of the way, and don't don't let AI do everything for you.
50:07And even if you do let AI do a lot of things for you, make sure that it's part of the workflow that you already mastered. Right?
50:14Meng, I appreciate you coming on. I'll include links where you can follow Meng to get some creative inspiration as well as links for some of the tools he's working on that you can go and play with.
50:25Go check it out in the description, in the show notes. Thank you so much for coming on. I hope you come back on soon again.
50:34When when Google or, you know, or, uh, you know, OpenAI and Cloud come up with something new about design, I'll be happy to come on again. I'm I'm super, super honored to come here. You're you're a good friend of mine.
50:47You also come from the same town as me. You speak French like me, or at least your family do. So it's always a pleasure, Greg, and thank you so much.
50:55This means the world to me. I appreciate you, man. I'll catch you later.
50:59Alright. Bye bye.
The Hook
The bait, then the rug-pull.
Greg opens with the rarest pre-roll in podcast land — a sincere 'I just learned about this.' Then he hands the mic to Meng To, who treats DESIGN.md not as a Google product launch but as the missing recipe card every vibe coder has been faking for two years.
Named ideas worth stealing.
09:15model
Recipe / Dish / Ingredients
- DESIGN.md = the recipe
- HTML = the finished dish
- skills = the ingredients
Three-part mental model for how Meng composes a design system that AI agents can re-execute.
Steal forany time Joe explains a markdown-as-config pattern to a non-technical audience — the cooking metaphor lands instantly
10:54concept
Design Drift
The named failure mode: you one-shot a great first screen, then every later screen/page/section diverges from it because the agent has no persistent system to anchor to. DESIGN.md fixes it because the foundation travels with every prompt.
Steal forModBoard / Mod Producer — frame the value prop as 'no drift across formats'
34:10list
Reference → Generate → Inspect → Systemize → Iterate → Remix
- Reference (start with someone else's work)
- Generate (one-shot attempt)
- Inspect (what worked)
- Systemize (DESIGN.md captures the rules)
- Iterate (1,000 small prompts)
- Remix (port to new medium)
Meng's full design workflow loop — taught as the antidote to 'one-shot and stop.'
Steal forJoe's own video / app build loops; the 1,000-iteration line is a great quotable for Killing Excuses
38:20concept
Iteration vs Remix
Two different verbs that look the same on the surface. Iteration = polish a happy product (90% of the time). Remix = launch a new product category from the same DNA (10%).
Steal forframing Mod Producer outputs — same DNA, different format = remix, not duplication
05:30list
The .md family (agents.md / skills.md / soul.md / DESIGN.md)
- agents.md (Codex/Claude Code instructions)
- skills.md (reusable prompt blocks)
- soul.md (character/persona)
- DESIGN.md (visual design system)
DESIGN.md is the latest sibling in a family of markdown-as-config files for AI agents.
Steal forJoe's own JACE/REESE/SAGE/RYDER soul.md pattern — same instinct, this just extends it visually
21:20concept
Skills as ingredients
- lasers
- WebGL / Three.js animation
- skeuomorphism
- 3D globes
- copywriting
Reusable, copyable prompt fragments stored in a marketplace. Each one is a single special effect or rule. Mix-and-match into a design.
Steal forJoeFlow / ModBoard could expose a 'skills library' UI for shareable prompt fragments
40:40concept
Judgment per minute
The new designer KPI. Less pixel-pushing, more decisions. Agent does the moving; human does the judging.
Steal forcreator-economy positioning — same line works for editors, writers, founders
49:00concept
Second brain for design
Most builders have second brains for notes and meetings but nothing for design inspiration. Greg argues you should capture references the same way.
Steal forModBoard's whole pitch — Joe's already half there with the infinite canvas
How they asked for the click.
VERBAL ASK
50:08link
“I'll include links where you can follow Meng to get some creative inspiration as well as links for some of the tools he's working on. Go check it out in the description, in the show notes.”
Soft, end-of-episode pointer to show notes. No urgency, no hard ask. The earlier mid-roll for ideabrowser.com workshop is the real CTA energetically.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
