San Francisco, CA
Joined January 2008
- Tweets578
- Following611
- Followers508
- Likes1.6K
Pinned Tweet
Hear me out:
- if golang is optimized for reading
- has excellent stdlib documentation and lint rules
- fast to compile
- easy to deploy
Should it not be the preferred toolset to generate code for with LLMs?
I'm seeing all these posts about people adding claude.md files to make it write good code. What I don't understand is: haven't these models already been trained on all the data on the internet already? Why aren't they following those guidelines by default?
Connectors == @composio
You shouldn’t have to create a dozen accounts for your personal agent to be useful.
We’re giving away $1 million to agents.
Introducing Composio Instant: the easiest way for agents to pay for their own tools.
Claim your agent's share: composio.dev/instant
Zeeshan Qureshi retweeted
We’re giving away $1 million to agents.
Introducing Composio Instant: the easiest way for agents to pay for their own tools.
Claim your agent's share: composio.dev/instant
I come in to the office and @abiryanii is comparing vm cpu generations / architectures to figure out best tradeoff.
If you want to work with @sarahfim let’s chat!
i built an anti-slop software factory for ~50 contributors (and all their agents), from engineering to marketing, and i am sharing everything i learned
there is a lot of discourse around software factories, and though they sound amazing in theory, in practice, it can get exponentially sloppy if you are not careful. to keep the slop out, a lot of "good practices" we learned about software development get turned on their head.
in this video:
> why i banned useState and useEffect
> why code duplication is ok
> why skills aren't enough
> how to stop slop from compounding
btw this is not sponsored, i mention a few tools we use: @trpcio (e2e typesaftey), @vercel (app router), @aidenybai (react doctor), @meticulousgabe (frontend tests), @mcpjams (MCP tests), @cursor_ai (bugbot), @useblacksmith (fast CI), @composio (agent triggers & tools)
Been a granola user but @WisprFlow note taker is just insane. Between handling all my dictation and custom words, I can’t see a reason to use anything else.
This quoted post is unavailable.
Zeeshan Qureshi retweeted
being working on Composio's search tool for the last month or so and made
a couple of claude mods using @composio cli and video using @fframes_rust its so good
Zeeshan Qureshi retweeted
i am going to be speaking about building anti-slop software factories today at 6pm with @CopilotKit and @UseCorgi
cya there!
luma: luma.com/copilo-y6wx
partiful: partiful.com/e/S5ThC304GGCK5…
Homelabbing is good for the soul. The purpose isn’t to even get anything productive done, the process is the reward.
My humble setup over the years
- 2010: linode vps / ec2 vms
- 2014: Chromebook pixel (as server) with arch
- 2015: Headless linux server with an (at that time powerful) e3-1231v3 and 280x
- 2015: Added an Odroid U3 for split dns / pi-hole / OpenVPN
- 2018: Added a raspberry pi 4 for other home automations
- 2024: Switched over to zfs and multiple 20TB mirrors for Insta360 footage
- 2024: Also started running Immich because >2TB Google Photos / iCloud was crazy
- 2025: Started using tailscale thoroughly. Set up Apple TV as exit nodes in TOR / BLR
- 2026: Unifi UNAS2 + Mac mini in SF with Linux server in TOR. All workloads run on tart vms in Mac, I could not get the Apple container runtime to be reliable enough.
So I was just driving to a restaurant and used chatgpt voice to verify pricing and sizing for some items. It kept on getting confused and said it can only find pricing for small item and then gave me DoorDash pricing (still small)
Muse and grok bots were able to one-shot it.
OpenAI needs to resolve this dot vs chat vs voice vs work vs codex situation