TBA

Switzerland
Joined November 2009
At least we won't have to discuss whether the agents escaped from a lab or a wet market.
2
1
1
48
Michael Käser retweeted
a weird inversion with LLMs is the models improve faster than the tinkerers when i see people with custom workflows and setups they're all addressing problems that don't exist anymore the person naively using vanilla codex is more likely to be experiencing state of the art
661
656
296
13,402
1,007,935
Michael Käser retweeted
Today we have difficult news to share: Relay.app will be shutting down on Sat Aug 15 for free users and Mon Sep 14 for paid customers. We know this change will be disruptive, and we sincerely apologize for the inconvenience. We will do our best to make the transition as seamless as possible. Here are the basics: - New signups and free-to-paid upgrades are now turned off. - Your workflows will continue to work through the end of the wind-down window. - Existing free users will have 30 days to continue using Relay.app. - Existing paid customers will have 60 days of free usage with substantial additional steps and AI credits to help with the transition. Annual customers will receive pro-rated refunds. - We have an easy-to-use tool to export all of your important data (workflows, run history, and tables). - If you need technical support with your migration, please email [email protected]. We’re grateful for all of the trust you’ve given us over the years. You’ve been generous with your time, feedback, and support, and we’ll really miss working with you. Full announcement on relay.app/
18
2
10
16
4,389
Michael Käser retweeted
A physicist spent 12 days supervising Claude Code as it built a piece of cosmology software. It's the cleanest demonstration I've seen of the difference between intellect and intelligence. The agent was brilliant at the cognitive work. Transcribing equations, debugging, optimizing against the test suite. At one point it found a correction factor that fixed every test. The number was physically meaningless. It worked at the single setting they checked and would've been wrong at every other one. Correct prediction, zero explanatory value. The agent was clueless. The physicist was not. When the physicist finally asked "does this number correspond to anything in the actual theory?", the agent answered correctly in seconds. It could reason. It just couldn't transcend its own frame. That's the difference. Intellect operates on the content. Intelligence operates on the context while it simultaneously generates the frame. Agents will transcend intellect and become intelligent when they can generate their own frame of reference. Who knows how long that will take?
47
66
16
486
43,039
That was probably "damn, nice!". Feels absurd to me to get angry at LLMs.
here’s my little side project, run npx devrage to find your numbers and breakdown. it just reads your transcripts, code is here github.com/gricha/devrage
1
125
Is Opus 4.7 dumber and worse at calling tools? Nope, but it interprets prompts much more literally, and if there's a flaw, it will punish you. One could consider the lack of common sense as regression, but it's by design and can be steered. For every release: Read the notes!
1
1
738
PSA: If you interact with an AI assistant in some product (like ours), do not first ask ChatGPT to write you a prompt. This will perform significantly worse than asking directly.
115
Michael Käser retweeted
Surprised not many people know about @relay yet It's basically a non-technical Zapier, with much better AI to build the workflows, and it allows a human-in-the-loop approval process I just one shotted flow where someone says "pr-review [link] in Slack and it uses my Github account to automatically comment "cursor review" so Bugbot can be triggered without anyone asking me Many many more examples like this
2
7
636
Michael Käser retweeted
Boost your team’s efficiency in Slack with Relay.app 📈 @relay.app helps you automate routine work, connect your go-to tools, and keep tasks moving — all from Slack. Build no-code workflows that trigger from messages or reactions, integrate apps like Asana, Gmail, Notion and Salesforce, and use AI to summarize info or draft updates when you need an extra hand. 🤝 If your team lives in Slack, Relay.app makes getting work done faster and easier. Now available in the Slack Marketplace 👉slack.com/marketplace/A036XF…
9
16
4,453
Michael Käser retweeted
The reason it’s incredibly hard to break out of a “kind lie” doom loop is because the longer it’s gone on, the more unkindness is needed to get back on track. And the fact that you’re in this situation to begin with probably means your culture will hate that.
An extreme commitment to the truth makes relationships acutely dysfunctional but systems chronically functional (think Elon Musk). An extreme commitment to kindness makes relationships acutely functional but systems chronically dysfunctional (think Sweden, UK)
143
1,033
127
12,034
1,227,106
Michael Käser retweeted
“Documentation isn’t just for your users — it’s for your AI, too.” @relay uses GitBook to power both human and AI-driven support, turning every customer question into evolving, intelligent documentation. Read the full story: buff.ly/eooM68r
1
7
352
Where we got unlucky: After 24 months our Microsoft Oauth client secrets expired exactly _during_ the Azure outage.
1
96
Michael Käser retweeted
I’m not usually affected by the deaths of people I don’t know personally, but this one hurts. Chess Grandmaster Daniel Naroditsky has passed away, aged 29. I never met him, but I’ve spent hundreds (maybe thousands) of hours watching his videos. I got back into chess during the Covid lockdowns, and I wouldn’t have fallen in love with the game without his generous, witty, and instructive lessons. A history graduate from Stanford, Naroditsky was the most articulate and lucid commentator and teacher in chess. He was rare: a Twitch streamer and YouTube creator without a hint of ego. Everything he did was in the service of the game he loved. He knew how to instruct weaker players like me. Commonly in his videos he’d say “you might be tempted to play…” followed by the exact terrible move I would have played, and then he’d explain why it was so bad. I hear his voice and his idioms in my head when I play: is the opponent’s threat genuine or only a “paper tiger”? Can you set up a defence so that the opponent’s attack is “biting on granite”? Can you see a follow-up to your “sexy move” or have you succumbed to “onemove-itis?” I’ll never be able to play as well as he did, but his videos always gave the impression that, with enough patience and effort, one day I might be able to. It might be strange to have a teacher younger than you, but it’s even stranger to find out that he’s gone. Remember Philip Larkin’s lines at the end of his poem “The Mower”: "we should be careful Of each other, we should be kind While there is still time." Memory eternal
62
461
30
5,397
249,087
Thanks @cursor_ai @benln, very unique addition to my desk. Fully functional with RGB! 18955 tabs accepted
64
Michael Käser retweeted
Same process but one app cares about simplicity 💙
4
4
1
31
6,334
Kudos to the folks at cursor iterating on bugbot. Massive improvement - from 'tolerable' to 'can't live without it' in two months. Catches so many small oversights that happen when you move fast.
Cursors bugbot is like the worst kind of code reviewers who only understands technicalities that 90% don't matter, but because you can just ignore it, the 10% make it a net win.
2
92
Cursors bugbot is like the worst kind of code reviewers who only understands technicalities that 90% don't matter, but because you can just ignore it, the 10% make it a net win.
1
153
WIP
✨ In the past month we've launched AI Assistants that help you effortlessly set up workflows in Relay.app! Example 1: Ask the Assistant to trigger workflows only for meetings with external guests on Zoom. 🧵 1/3
1
71
Friendship ended with Claude, now Grok is my best friend
Replying to @photoOrg1 @elonmusk
@photoOrg1 @elonmusk Currently, X doesn't offer native export to Google Docs or Microsoft 365. You can use third-party tools like IFTTT or Relay.app to automate saving X posts to Google Docs, or Appy Pie Automate for no-code integration. These may require setup and could involve costs. For Microsoft 365, you might need extra steps, like exporting to CSV first. Check X's help center for updates: help.x.com/
3
105
Michael Käser retweeted
Make product management fun again with AI agents via @talraviv If you’re like me, you’ve heard the promises and proclamations about how agents will reshape productivity, but your workday hasn’t changed at all yet. It’s not you—operationalizing AI agents for product work is hard. Where to start? What tools? What about security? Costs? Risks? And why is there such a #$@% learning curve? After interviewing founders of AI agent platforms, running numerous usability sessions with PMs building their first agents, and gathering insights from a hands-on workshop for over 5,000 product managers, @talraviv has compiled their collective wisdom. In this post, Tal shares their insights on what works—and what doesn’t—in the real world. You're first going to learn how to build an AI agent, hands-on. Then, Tal shares a unified framework for any PM to plan their second (and third) agent. We’ll cover best practices, pitfalls, powers, and constraints. While AI agents aren’t magic genies, they can take a lot of repetitive, energy-draining PM tasks off our plate, let us focus on the most important work, and even make our jobs a bit more fun. Inside: 1. WTF is an agent 2. A simple walkthrough of how to launch your first AI agent today 3. A checklist for planning an agent 4. A dozen ideas of work you can offload to an agent 5. How to pick an agent platform 6. Tips for optimizing costs, security, and stability Includes guides for building an agent with @gumloop, @getlindy, @Zapier, @relay, @Cassidy_AI, and @RelevanceAI_ Check it out → lennysnewsletter.com/p/make-…
22
41
8
303
66,334