Gleb Stroganov

1.4K posts

Gleb Stroganov banner
Gleb Stroganov

Gleb Stroganov

@strongeron

Product Designer at @evilmartians • co-creator AgentPrism #Harmony • Vinyl Enthusiast • Music Festivals Fan

Lisbon, Portugal Katılım Ekim 2013
884 Takip Edilen371 Takipçiler
Sabitlenmiş Tweet
Gleb Stroganov
Gleb Stroganov@strongeron·
The worst part about the agents is that they fail silently. They loop endlessly, skip steps, and give wrong answers… It took me 4 hours to parse a JSON wall of agentic traces to find a fix. We built and open-sourced something that cuts AI agent debugging time to 30 seconds.
English
1
4
18
3.1K
Gleb Stroganov
Gleb Stroganov@strongeron·
Amazing work and perspective with chrono-grid breakdown. The part that stops me: each edition isn't a new poster, it's a new system — rules that turn match data into form. This is generative design with actual taste.
Zeh Fernandes@zehf

Many people asked how the World Cup posters are generated. I wrote a breakdown of the system behind them. It covers the chrono-grid, visual encoding, and the rules that transform a match into a poster. zehfernandes.com/posts/how-i-tu…

English
0
0
1
147
Gleb Stroganov
Gleb Stroganov@strongeron·
The honest reason I built this: I kept opening agent-coded repos and couldn't tell which components were real and which were dead ends abandoned three prompts ago. So the audit became a set of skills – Storybook Workbench. Point it at the mess → get the map and act.
Evil Martians@evilmartians

An agent can spin up a web app in an afternoon. But which of those 469 component files actually matter, and which are dead ends abandoned three prompts ago? That's why we built Storybook Workbench: a set of agent skills that audit your AI-generated UI and lay it out in Storybook. There, you can see what are “real" vs slop components, view how each component appears on different screens, and get a design system health score. @strongeron, the author, says this works for when you inherit an agent-coded prototype or move a design system into code. Use it for the first pass and every iteration after. Read the full story below.

English
0
3
10
799
Gleb Stroganov
Gleb Stroganov@strongeron·
So true. I see myself now contributing across 4 of these points. The most vital one — knowing when to stop at one stage and transition to the next. A strong prototyper can get stuck iterating on ideas forever; a builder might resist handing off to a grower. The meta-skill is stage awareness — knowing what the product needs now, not just what you're good at.
English
0
1
1
135
Boris Cherny
Boris Cherny@bcherny·
As engineering, product, design, DS, etc. melt into a new kind of role, I was reflecting on what roles might look like in the future. For example, when I look at the Claude Code team I see what I think is five archetypes: 1. Prototyper: comes up with brand new ideas; churns out many ideas, most of which don't ship 2. Builder: quickly turns a prototype/idea into production-grade product/infra 3. Sweeper: cleans up the UI, simplifies the code and system, unships, optimizes performance 4. Grower: takes a product that has been built and iterates on it to improve Product-Market Fit 5. Maintainer: owns a mature system to make it secure, reliable, fast, and efficient as it scales Many people span across 2 roles, and sometimes 3 roles. I also notice that these roles are not really tied to job function -- eg. across Anthropic, some designers match category 1, some 2, some 3; same for engineers, PM, DS. A healthy team needs a mix of these, depending on the product: - A product that is new and pre-PMF needs people that are strong at 1+2+3 - A product that is growing and has found PMF needs 2+3+4 and some 5 - A product that has strong PMF needs 3+4+5 and some 2 Maybe product roles of the future will look more like this, and less like the domain-specific roles of today?
English
887
2.4K
19.8K
3.1M
Gleb Stroganov
Gleb Stroganov@strongeron·
When an agent run finishes, I don't care about the green checkmarks. I care about the one or two steps that quietly went sideways. That's the real work of building agents: not running them, seeing where to tweak. We shipped n8n.io support in AgentPrism. Bring your hardest cases.
Kirill Yakovenko@KSYakovenko

Just shipped: n8n.io integration for AgentPrism, our open-source AI observability project. Better than n8n's built-in trace viz? We store logs forever, or at least until we run out of space. Beyond that, no clue. Closed beta: agent-prism.evilmartians.io #n8n

English
0
0
3
118
Gleb Stroganov
Gleb Stroganov@strongeron·
My design log lives in Storybook. So does my final call. Every option I'm weighing sits there as a living record — this card and its "Coming Soon" badge, checked against dark and light photos, all pulling real tokens from our design system. They line up side by side in the environment they'll actually ship in, so weeks later I can compare them for real instead of from memory. When one option wins, an agent applies it across the app in minutes. Explore, decide, ship — all in one place. That's the part of founding design I love most.
English
0
1
7
445
Gleb Stroganov
Gleb Stroganov@strongeron·
A feedback survey is only as good as the moment it interrupts you. So I don't design that moment in a mockup — I tune it in Storybook. Same widget, pulled out of the app but running in a real environment, where I can tweak the scroll trigger and the show/hide behavior until it feels invisible. Two variants of the same survey here: one hides on scroll down, one on scroll up. Watching them side by side is how you feel which is right.
English
0
0
0
78
Gleb Stroganov
Gleb Stroganov@strongeron·
Bridging design ↔ code the AI-native way. We built an environment where designer + backend can ship together, compound, and create skills. Design meets code in @storybookjs: our single source of truth for auditing, iterating and shipping. The loop finally closes. Never been this productive. Full story below 👇
Evil Martians@evilmartians

A designer and engineer worked together to ship a production-ready MVP in four weeks using Rails + Inertia. In this post we share how we built @thicket, our agentic coding stack, the skills we built, and what worked: evilmartians.com/chronicles/2-m…

English
0
0
1
121
Gleb Stroganov
Gleb Stroganov@strongeron·
@stephenhaney So cool! Congratulations on the release. Explorations with modifiers my favourite beat.
English
0
0
1
30
Stephen Haney
Stephen Haney@stephenhaney·
Good morning! The Pen Tool is live in Paper Draw shapes, generate SVGs, or vectorize images, and edit the points and paths. Paper uses real SVGs so exports are 1:1, no surprises. The new SVG workflow: create → refine → ship. Try it out! And show us what you draw.
English
94
66
1.3K
4.1M
Gleb Stroganov
Gleb Stroganov@strongeron·
Our @thicket project: Validation is infrastructure we built to ship with confidence.
Evil Martians@evilmartians

Building a product is one thing; knowing someone is willing to pay for it is an entirely different game. After shipping @Thicket's marketplace in 2 weeks, our team built a full demand validation infrastructure around it. This resulted in: - 36K impressions - 128 cold-traffic waitlist joins - A landing page variant confirmed at 95% statistical confidence ... and all on a $2K ad budget. In this post, a repeatable validation playbook with sequenced tools, just a few rules, and founder-operable: evilmartians.com/chronicles/fro…

English
0
0
3
246
Gleb Stroganov
Gleb Stroganov@strongeron·
Meet my work, my projects, and the things I pretend are just hobbies.
Evil Martians@evilmartians

Meet Gleb (@strongeron), one of our product designers who's spent over 10 years learning how to design better developer tools. As soon as vibe coding became a thing, Gleb was one of the first designers to try it out internally and has now built multiple production-ready products. He's also a music lover, collects vinyl, enjoys live performances, and plays music (from bass guitar to modular synths). If there's a festival in town, you can be sure Gleb will be there, and in between performances, he'll be digging for new vinyl records. See his work: evilmartians.com/martians/gleb-…

English
1
0
8
168
Gleb Stroganov
Gleb Stroganov@strongeron·
@stephenhaney Congratulations on the release! What a huge milestone. More ideas to explore and polish ahead!
English
1
0
1
357
Stephen Haney
Stephen Haney@stephenhaney·
Hello! Today we're releasing Paper Desktop Paper is now a canvas for Cursor, Claude Code, Codex. Any agent can read and write html to Paper. • push or pull from your codebase • pull real data from anywhere • less work, more design What will you ship? Sound on 🎶
English
357
389
5.8K
1.7M
Gleb Stroganov
Gleb Stroganov@strongeron·
@joshpuckett It’s an impressive piece of work across every required angle—refreshing knowledge, enabling learning, and inspiring exploration. Thank you for creating and sharing this craft!
English
0
0
1
90
joshpuckett
joshpuckett@joshpuckett·
Interface Craft is now open. It’s a growing library of resources for those who are committed to crafting experiences and interfaces with uncommon care. I hope you’ll consider joining:
English
161
155
3.3K
509.5K
Gleb Stroganov
Gleb Stroganov@strongeron·
That set of skills + agents + compound approach = effective, focused AI work. Spend time understanding the idea, try it in your flow, and the black box becomes your personal compound engine of ideas + solutions – unlocking AI across any task.
Every 📧@every

We just published the complete guide to compound engineering— @kieranklaassen's approach to AI-native development that has over 7,000 GitHub stars. Drop it straight into Claude or ChatGPT and start using it: every.to/guides/compoun…

English
1
1
1
184
Gleb Stroganov
Gleb Stroganov@strongeron·
We're helping teams debug AI agent logs with intuitive visualization. Looking for early adopters facing similar challenges—read our story and reach out if you're debugging complex agent behaviors. Let's solve this together.
Evil Martians@evilmartians

We're looking for early adopters to explore and help us improve AgentPrism, our open source library of React components for visualizing traces from AI agents. Check this post for more info and GitHub link: evilmartians.com/chronicles/deb…

English
0
1
3
211
Gleb Stroganov
Gleb Stroganov@strongeron·
Debugging agents shouldn't feel like archaeology. Built AgentPrism with @QuotientAI and @JuliaANeagu: open-source React components that turn complex traces into clear diagrams in a fraction of the time. Our full story from concept to production 👇
Evil Martians@evilmartians

Agentic traces contain perfect information about an agent’s behavior with every plan, action, and retry. But that information gets lost in a sea of JSON. So we built AgentPrism: open source React components that turn traces into visual diagrams for debugging AI agents. You can plug in your OpenTelemetry data and see your agent’s process unfold: messages, tool calls, retries. @QuotientAI automatically monitors, analyzes, and improves AI agents. Being able to review traces quickly is paramount for their research, and for their customers. “Dealing with agent traces was one of the biggest frustrations for our researchers and a huge time sink. All of that has gone away since adding AgentPrism, and we’re excited to bring that functionality to our users.” — @JuliaANeagu

English
0
0
5
296
Evil Martians
Evil Martians@evilmartians·
Agentic traces contain perfect information about an agent’s behavior with every plan, action, and retry. But that information gets lost in a sea of JSON. So we built AgentPrism: open source React components that turn traces into visual diagrams for debugging AI agents. You can plug in your OpenTelemetry data and see your agent’s process unfold: messages, tool calls, retries. @QuotientAI automatically monitors, analyzes, and improves AI agents. Being able to review traces quickly is paramount for their research, and for their customers. “Dealing with agent traces was one of the biggest frustrations for our researchers and a huge time sink. All of that has gone away since adding AgentPrism, and we’re excited to bring that functionality to our users.” — @JuliaANeagu
English
17
75
812
101.4K

Keşfet

@storybookjs @stephenhaney @thicket @antiflasher @joshpuckett @claudeai @elonmusk @BarackObama