Sai Kota

plumber @8090_factory

[Sai Kota @saikotaaa]

@razeenali01 my running mileage

[Sai Kota @saikotaaa]

@DominikTornow @Cloudflare 🤯🤯

[Sai Kota @saikotaaa]

@razeenali01 Read your own information. Make your own conclusions. Don’t delegate your understanding

[Illiquid Insights @IlliquidInsight]

Data centers are being funded through SPV structures. Hyperscalers lease the facilities, and often guarantee the debt and provide equity.

This allows Big Tech to keep the debt off balance sheet.

[Sai Kota @saikotaaa]

@tpschmidt_ lol I just fixed this issue in our deployment yesterday

[Sai Kota @saikotaaa]

Crypto - local client & remote node.
Agents are converging to similar architecture. If we are going to use agents for increasingly ambitious tasks, they will take time to run, so we don’t need to babysit them locally

[Sai Kota @saikotaaa]

@dagger_io @bot how is this different from @NxDevTools

[Sai Kota @saikotaaa]

@yucjonathan did this at @8090_Factory 3 months ago. Man is a visionary

[Tak 🦞 @cherry_mx_reds]

Anthropic is now recommending that companies REMOVE unnecessary subagents from their AI systems.

This is a pretty significant shift in agent architecture, with their team finding that a single agent equipped with skills can frequently outperform complex multi-agent systems, and one customer seeing roughly 2x faster execution.

The emerging rules of thumb:

→ Skills: Teach the agent HOW to do something, including procedures, expertise, and instructions.

→ Tools / MCP: Give the agent the ability to DO things through APIs, actions, and live data.

→ Subagents: Use for parallel work, isolated context, or tasks a single agent struggles with.

The interesting part is progressive disclosure. Instead of spawning specialized agents or stuffing everything into a giant prompt, one agent discovers and loads skills and tools as needed.

Their advice is to fold existing subagents into the main loop, turn their instructions into skills, expose their actions as tools, and run evals to see if performance holds up.

As models have gotten smarter, improving agent architecture is becoming less about what you can add and more about what you can delete.

https://t.co/DNq60Skkmc

[Sai Kota @saikotaaa]

This would become blueprints in @8090_Factory
A well-defined system w clearly defined properties/invariants/interfaces/protocols can be rewritten a million times. It can tested a billion times in a trillion different ways

[Grigore Rosu @RosuGrigore]

1/4 etcd (@etcdio) is the data store behind Kubernetes. We wrote 157 properties it must always satisfy, from its docs, comments and code. Then we searched for runs that break them. Three high-impact issues: one disclosed privately, two below.

cc @serathius @CloudNativeFdn

[Sai Kota @saikotaaa]

@razeenali01 @sojoodi

[Sai Kota @saikotaaa]

@threepointone @dillon_mulroy me too.. and idk either of you!! Wtf

[Naval @naval]

Networking is overrated. Become first and foremost a person of value and the network will be available whenever you need it.

[Sai Kota @saikotaaa]

Please don’t do this 😭😭
Just use hooks + @langfuse

[Ben Holmes @BHolmesDev]

I met a team of devs regularly zipping up Claude and Codex sessions from their computers and shipping them to a shared S3 bucket.

Why? Because collecting data on agent convos is that valuable. I think shared cloud sandboxes are the way though to avoid all the zipping 😄

[Chamath Palihapitiya @chamath]

Founders send me pitches every day. Here’s a better way: come to our hackathon!

EVOLVE: 20 hours, overnight at 8090 in Redwood City, Nov 13–14.

Build AI agents and tools that build software. The winning team, beyond recognition and a cool trophy gets two 1:1s with me, to help you refine a pitch or idea and then a follow-up.

Use the time however you want. Ask me for advice, ask me for money or just make fun of me. All allowed if you win!

Apply: https://t.co/zvRSjRh3bW

[Sai Kota @saikotaaa]

@AfifaSiddiqua_ leave

[Sai Kota @saikotaaa]

agents remove middlemen. go to the source you brave soul

[BOOTOSHI 👑 @KingBootoshi]

if you got fucked by Railway I highly recommend just using AWS and having Opus 5.5 handle dev ops with Terraform/OpenTofu

it seems intimidating at first but it’s actually incredibly easy now

you can even have your agent handle this via TDD first so it’s guaranteed secure!

[Will Stewart @NorthflankWill]

BYOC. Production Sandboxing. Enveloped Encryption. WAF. Runtime Scanners. Audit logs. Red teaming agents. Workload identity. mTLS. Automatic secret rotation. Network segmentation. Authenticated Network Endpoints. Deploy security fixes in 60 minutes. Automated Rollbacks. Activity Monitoring.

What am I missing?

[Sai Kota @saikotaaa]

not a sexy opinion but GitHub is cooking

[GitHub Next @GitHubNext]

How do you _guarantee_ your agent's code is correct? You prove it.

Introducing Proof Driven Development Kit (PDD Kit) from @MSFTResearch.

PDD Kit helps turn your plan into a formal spec, then proves the agent's code meets that spec for every input. No vibe verification. PDD Kit is integrated with @GitHubNext's Chopin, so you can plan as a team.

PDD Kit takes an app's login flow from intent to a verified pull request:

It discusses your code, infers your intent as it goes, and proposes a formal spec in Verus, a proof language for Rust.

Once you approve the spec, the Verus verifier automatically proves the code meets it.

Where proof isn't possible, it falls back to empirical verification with tests and computer-use agents (CUAs).

It works on Rust today. No new language, no change in how you produce the underlying code, making formal methods more accessible.

The goal is a verified software factory, where engineers review specifications instead of code.

Come see it in action at GitHub Universe, Oct 28–29. Swing by the Speakeasy and ask for GitHub Next!

[Sai Kota @saikotaaa]

logical replication is a fickle beast. WAL size can blow out when you ALTER column data types as well. The table update & large WAL kills compute. On top of that until subscriber consumes those WAL files your primary keeps burning through storage

[Sam Lambert @samlambert]

"In larger environments we've seen hundreds of gigabytes of WAL written by a single transaction"

[Sai Kota @saikotaaa]

@razeenali01 Auth man