고정된 트윗
AI tools keep getting better. But every session starts from zero. Your agent doesn't remember what the client rejected, what your design principles are, or the research behind the strategy. You re-explain everything, every time.
English
Christian Marc Schmidt
1.2K posts
Christian Marc Schmidt
@cms_
Founding Partner @schemadesign. Exploring the edge of AI, spatial computing, & data. Formerly @pentagram, @ideo, @microsoft. Author of https://t.co/xwg1kshEIy
Seattle, WA 가입일 Mart 2009
1.7K 팔로잉789 팔로워
@joshpuckett @frank_chimero @benjitaylor This is the right wedge. Designer tools in the browser, not just developer tools. Excited to try the alpha.
English
@frank_chimero @benjitaylor It’s coming! First alpha aiming for next week
joshpuckett@joshpuckett
Meet InterfaceKit, a design tool for working with agents. Refine components live in your app → click to copy prompt. Your agent gets the element data, Tailwind classes, and Motion animations it needs to make your changes in one shot. No more guess and check.
English
Browsers have developer tools. They need designer tools, too. Would be cool if someone made a browser plugin to visually manipulate the DOM with familiar handles. Waiting for that a ha “Firebug” moment. Agentation from @benjitaylor is a great first approach at this.
English
This is the workflow. I just ran this exact setup tonight and the unlock wasn't the model or Paper (both incredible). It was feeding Claude the meeting notes from this morning's design review. It picked up interaction patterns we discussed 6 hours ago and generated 8 wireframes that reflected actual project decisions, not generic UI. The missing piece for most teams is structured context.
English
@shuding The restraint of keeping this at 5KB is the real story. Beautiful work.
English
Every project your team ships makes the next one easier. Not because people remember, but because the knowledge graph does. The left side is month one. The right side is month twelve.
English
@kepano Knowing when not to reach for AI is itself a design decision. Rules for structure, models for intent. This is the right call.
English
I have been working on Obsidian Reader for a over a year. I didn't want to share it until I felt it was good enough. It's finally there.
Consistent formatting for any article. Outline, syntax highlighting, nice footnotes, adjustable typography.
Runs locally. Just rules, no AI.
English
@jamesgaynor @nyk_builderz Will do. I'll DM you when it's ready.
English
Design engineering skills for agents are a good start. The next step is the organizational layer underneath. Your component conventions, your naming patterns, your past decisions about when to use animation vs. static transitions. That context lives in the team, not the blog. The skill file becomes a lot more powerful when it can reference a structured history of what you've actually shipped. x.com/emilkowalski/s…
English
@jamesgaynor @nyk_builderz It's not. The core ideas hold (structured markdown, wikilinks, frontmatter as schema) but you need real entity validation, permissioned collections, and conflict resolution before it works across a team. That's the thing we're building now.
English
@cms_ @nyk_builderz I'm actively thinking about this from a team perspective too. I'm not sure Obsidian is the right tool for an enterprise with even 50+ people. Is anyone tackling this at a larger scale?
English
Most teams store what they know in documents. Meeting notes, project briefs, strategy decks. The AI searches them and returns chunks of text. It works, but the AI is doing retrieval, not reasoning.
The shift is storing knowledge as typed entities with explicit relationships. A project links to its client, its team, its decisions, its research. An agent can traverse from a brief to the design principles that informed it to the client feedback that changed it.
Documents give you search. A graph gives you understanding.
English
This is the coordination layer agents have been missing. Tried it this weekend. The next unlock is shared organizational memory with agents that know what the company knows, not just what they're told to do.
dotta@dotta
We just open-sourced Paperclip: the orchestration layer for zero-human companies It's everything you need to run an autonomous business: org charts, goal alignment, task ownership, budgets, agent templates Just run `npx paperclipai onboard` github.com/paperclipai/pa… More 👇
English
@garrytan We run an entire design firm this way. All in markdown and using Git, from project history and client context to design principles and records of decisions made. The agents get smarter the more you add.
English
I also believe this: PM and design needs to be markdown and image files in git repos and developed using Claude Code alongside plan files and QA markdown test cases
Vaclav Milizé@clwdbot
the underrated consequence: if markdown is code, then product specs, agent instructions, and system prompts all belong in version control with code review. most teams still pass these around in Notion docs. the ones treating .md files like source code (with PRs, diffs, and CI validation) are shipping 2x faster because the AI and the spec evolve together.
English
We went with markdown over a database. The graph is frontmatter + wikilinks, Git-backed. Probably won't scale to millions of records but for an org-sized knowledge graph it works because LLMs read it natively, no serialization needed. Semantic search also works better when the source is already natural language. Still early though. What are you exploring?
English
@cms_ @Scobleizer @obsdmd @noahvnct What r your thoughts on databases? I’m thinking about this a lot btw! :)
English
A lot of the highly techie AI people in San Francisco use @obsdmd to store everything about their life and dreams. Er projects they are building.
@noahvnct laid out why better than I have seen.
Noah Vincent@noahvnct
English
Give an AI your project files and it generates decent work.
Give it a structured graph of your clients, projects, decisions, and team relationships, and it starts doing things you didn't ask for.
It catches when your new scope contradicts what you told the same client last quarter. It pulls up a reference from a past project that nobody on the team thought to look for.
The difference is memory.
English
Notion is doing interesting work here. The distinction for me is between giving AI access to documents and giving it a graph of typed entities with explicit relationships it can traverse. Both compound but differently. Documents give you retrieval, while a structured graph gives you reasoning across connections.
English
@cms_ @Scobleizer @obsdmd @noahvnct This is most of the AI notes apps!
Notion does the same thing. It’s a graph.
English
The problem is continuity. We've been running a structured knowledge graph to operate our design studio for the past year. Every project, decision, or deliverable feeds the same system. The agent starts from what the team already knows. That's when the work actually compounds.
English
AI tools keep getting better. But every session starts from zero. Your agent doesn't remember what the client rejected, what your design principles are, or the research behind the strategy. You re-explain everything, every time.
English
@nyk_builderz The graph is Git-backed, so version control handles the mechanics. The harder problem is semantic, for example two people describing the same concept differently without realizing it. That's where matching against existing entities before anything gets committed is key.
English
Obsidian's real strength is that it provides a portable structure that AI can reason over: markdown, wikilinks, and YAML. This creates a graph, not just a collection of notes.
The single-player problem is real, but the answer isn't going back to Notion. It's bringing that same structured graph to teams without losing what makes it work for agents.
English
@Scobleizer @obsdmd @noahvnct Obsidian is good but only for single player.
Notion is best for teams/cloud.
English