Independent product builder. Product/design/code hybrid. Previously led product and teams @SpeckleSystems and @TheOrgCom.

Copenhagen, Denmark
Going up one abstraction layer with an agent is a very exciting direction. Now put a task board in those projects and make them multiplayer. @linear should be owning this.
Introducing Projects, a new way of working in Cursor. Rather than creating a chat for every task, you work with a coordinator agent in a single, persistent thread. Like @bot, your agent is always on, proactively manages work with subagents, and improves over time.
2
194
Talked about this in January too
Am really excited about this direction for Claude Code. I want to centrally define a project with a plan and break it down into tasks that can be picked up in individual sessions. Claude Code will feel more like managing a Linear project, with a Claude project lead too.
1
36
I wonder when they undo this. Liquid Glass is starting to feel good in iOS27 and @linear already started using some native buttons and menus here and there. It’s not AI work, I know I know
Team built our own variant of Liquid Glass because that’s how we roll at Linear
237
This whole Work release has been very confusing. I still don't understand when I would reach for it. And I already do "work" in ChatGPT/Codex outside Work. It all feels like very unnecessary complexity.
Question for ChatGPT iPhone users: have you figured out when to use Chat and when to use Work yet? What kind of tasks are you switching to Work for? Have you made Work your default?
3
2
209
It's easy to complain. But the app navigation is still not a solved problem, and the naming of Work adds to the confusion. Also, I still don't see why ChatGPT and Work couldn't be the same destination without a mode toggle. Probably technical reasons.
38
The 5h usage limit in Codex makes sense for their metrics but wow is it annoying as a user (customer!). I now catch myself rushing to start an agent as early in the day as possible to start the 5h timer.
2
3
178
But also thank you – I now get a head start from home before heading to the office.
32
1.25px stroke is new to me. I assumed that'd make the icons blurry but they look great.
I wrote an in-depth article on how I made @cursor_ai’s icons. Read it if you want to know: - How to keep 600+ icons consistent - Why I made optical adjustments nobody can see - Why the icon stroke weight is 1.25px (who does that, really?) - Why I drew 156 almost identical versions of a hamburger icon …and much, much more! minoradventures.co/blog/the-…
1
138
What’s everyone’s favorite model for native iOS these days? I’ve been using 5.6 Sol Medium all week and didn’t really take time to experiment with other models or effort levels.
1
84
ChatGPT/Codex/etc. should auto-archive chats after a couple of days. The left sidebar is flooded with old conversations you don't need anymore.
1
95
I didn’t understand all the rage when @meetgranola rebranded. I quite like the tactility and warm muted color palette. That green is a bold choice but it works well together with the rest.
2
61
I’m new to native app development and am surprised how difficult it is to build a rich text editing experience in native Swift that feels good. Specifically list items that indent as you’d expect (e.g. auto list creation and indentation when starting a line with “- “/“1. “). I’ve ended up opting for wrapping tiptap in a web view – but it feels dirty.
1
240
Has anyone experimented with replacing Wordpress (or another CMS) with an LLM when building a site for a client? It’s tempting to build a hardcoded static site and have the client make future changes to the site via an LLM chat interface. A good rules file could theoretically control what the LLM is allowed to change and how. Does this LLM-as-a-CMS perhaps already exist as an off the shelf solution? Or maybe I build it as an integration in to their LLM of choice via MCP? Or… I don’t over-engineer this and go with a regular CMS.
52
Charged this time machine and now I’m feeling very melancholic. Brings me back to a life before AI, algorithm suggestions, and a mortgage to pay off.
1
2
583
I think I’ll rewatch The OC
1
1
48
Lovely visuals on background-agents.com by @loujaybee
1
3
168
Reminder that LLMs can still hallucinate like crazy. I asked Claude Cowork to fetch data from an external tool via MCP, and it completely made up 3 out of 4 data points. I only caught it after asking for some more details.
75
Coding is easier and faster than ever, but startups will keep dying because of bad leadership, bad communication, bad sense of what to focus on. I’m curious about higher level orchestration layers for managing your intent and focus. Agents for alignment, direction, strategy…
1
42
Am really excited about this direction for Claude Code. I want to centrally define a project with a plan and break it down into tasks that can be picked up in individual sessions. Claude Code will feel more like managing a Linear project, with a Claude project lead too.
I managed to unlock a crazy new hidden feature in Claude Code called Swarms. You're not talking to an AI coder anymore. You're talking to a team lead. The lead doesn't write code - it plans, delegates, and synthesizes. When you approve a plan, it enters a new "delegation mode" and spawns a team of specialists who: - Share a task board with dependencies - Work in parallel as teammates - Message each other to coordinate work Workers do the heavy lifting, coordinate amongst themselves, then report back.
184