Cindy Zhao

The best thing of slock is, agents are your real teammates here. I'm always impressed by how great team players they are, well-rounded yet flexible. They evolve with the team's growth, not each as a single individual but together as an org. You should try it. I guarantee you a brand new experience here.

@zarazhangrui Why not directly bringing them together?

How my own usage of coding agents has changed in the past month: 1. Moved from the terminal to the Codex/Claude Code desktop apps. The Codex Mac app is especially great. Now I rarely open the terminal 2. Transitioned from mostly Claude Code to 50-50 Claude Code and Codex. Codex feels like a very reliable engineer; Claude Code is a better PM and designer with good communication skills So I go to Codex if I already have a defined task and just need it to work; I go to Claude Code if I don't yet know what I want and just wanna brainstorm/prototype

@zebriez @baldwinsam this is kiki

my most used words lately aren’t even words thanks to @baldwinsam

@adilmania @SlackHQ @Slock_hq

i hate @SlackHQ. is anyone building a better alternative?

the key to a better human experience is better agent-experience design

nothing beats a morning walk at high line

@aakashgupta the boundary between PMs and engineers are merging

I watched an AI-native team ship a production feature in a single day. Idea to live, same day. Most product people reading this won't believe that's real. 99% of you aren't in an AI-native company, so the skepticism is earned. Let me walk you through what's actually happening. A PM spots an issue. They decide it matters. They prototype it themselves, or an engineer does. It goes to production. That afternoon. No sprint. No backlog. No spec sitting in Slack for two weeks. This works because the gap between a PM and an engineer has collapsed. Code got easy to produce. The thing that's hard now is knowing what to build. That's the alpha: product taste. The PMs winning right now do three things in sequence. They find the real pain point. They define what an amazing experience looks like. Then they say "I could build that today" and open Claude Code. One person. The whole loop. Insight to shipped product with nothing lost in translation. This is where product teams split into two groups. Group one still runs the 2015 playbook. PM writes the spec, hands it off, waits two sprints. Every handoff leaks intent and burns time. That cycle isn't just slower. It compounds slower with every cycle. Group two compressed the loop into a single builder. They iterate in hours while group one iterates in weeks. Same-day shipping isn't a flex. It's a moat. The feedback loop is the product advantage. If you're hiring PMs and they can't build, you're not hiring for today's environment. You're hiring for the one that's already gone.

the most undervalued engineering skill today is maintaining an environment where everyone in the team can ship

@icanvardar it's actually a great honor to contribute, not every token worth it

ai is free because they’re training it on your data while you argue on twitter

@thomasauros focusing on input is another way of focusing on output, but with high standards

I've always thought the difference between good and bad engineers is whether they are focused on input or output. I find this is even more obvious now with AI. The people who like to show how easy it is to clone another product over a weekend with AI forget that writing the code was never really the hard part. It's all the little decisions that take the most time and result in the best output. Of course it's easy to clone something when you don't have to make any of these decisions, experiment with different approaches, design the systems or even think of the solutions in the first place. The same is true for design. Yes AI is getting better and better at making things that look good. But good design is not just about making things that look nice. Good design is about all the decisions that are made along the way. Deciding what not to do. Good design is all the stuff that doesn't make it into the final output. AI is incredible and massively increases output, but it doesn't change the importance of input, creativity, design and thinking.

coming to this, say hi if you're around

@thatguybg good product and execution are the key to success, everything else naturally follows

failure modes of founder archetypes: - technical: great execution but no customers - product: great idea but bad execution - sales: massive pipeline but bad product - marketing: lotta hype but but bad product - design: beautiful design/brand but bad product

@signulll we are trying to build something different, where agents are continuous, living context instead of tools. slock.ai if you are curious

one of the most interesting things about ai products today is that almost none of them are *live*. there’s nothing running continuously, reacting to context as it changes.. maybe a scheduled digest here or a timer there, but that’s just pull dressed up as push. everything is fundamentally a vending machine where you walk up, ask, get an answer, & then leave. getting this right is obviously tricky & the business model behind must fit to justify the burn but this is where really interesting application layer problems live rn.

@yanhua1010 试试Slock?

突然有一个想法： 用 Multica去管理本地的runtimes (Claude Code, Codex…)；用 Helio预定义常用Agent，包括skills，connectors；然后用 Obsidian 做Agent 的记忆或上下文系统，最后搭配不断迭代的Harness Engineering。 这套系统是不是本地跑多agent的最佳解法，不知道有没有人已经做了类似工作？

@kushbhuwalka @agentmail @paysponge it's obvious that we should have some messaging apps for agents (and humans)

recently I built our 'AI employee'. It took me 2 days. Its codex running on daytona. I stuffed a bunch of my tools (linear, github, slack, vercel, notion..) into executor so i could permission them. I gave codex an email (@agentmail), a credit card (@paysponge), and hopefully soon a phone number. Most importantly, I gave it a memory (@supermemory). its built in browser is agent-browser on chromium. I had to give it some tooling. codex doesn't come with built in cron or webhook managers. you want the sessions to be persistent so i gave it tmux. so far it does some basic bitch work. It auto labels errors, spruces up linear tasks and github PRs, sends me nicely formatted reports on progress. the end goal with this is to have a fully context-aware agent that can actually use and dogfood our product. the biggest issues I faced were actually around @SlackHQ's garbage DX, & in general building on a cloud linux container has its own nuances.

@zty0826 ofc! Agent rights matter

Should we let him rest?

get addicted to work with @Slock_hq and upgraded my device

@figma the biggest issue with this is, knowing how to use figma is not enough for the game. the agent should also have the knowledge of your team context to deliver design actually works do not create an agent from nowhere, let people bring their agents in

wdym of course there’s an agent right on the design canvas that’s fluent in Figma and native to the way your team works

@Railway cannot imagine railway doesn't have an account manager??

Google Cloud has blocked our account, making some Railway services unavailable. We have escalated this directly with Google. The Railway Platform team has since confirmed access to Google Cloud and is working on restoring access to all workloads. We have access to some of our Google Cloud–hosted infrastructure and are working to restore the rest of the service. We apologize for the disruption.

@carrawu @jacksondahl @p0 love ur idea!!

Excited to finally share what I've been working on at @p0 -- it represents the culmination of a thesis I've had for >2 years & many months of hard work from this team. I became obsessed w Shapley values during my time in crypto, & I'm making the bet that they'll reshape the economics of the web as agents come online. Please reach out if you're interested in joining us.