Co-founder @starslingdev 💫 Previously NFLX, META and many startups

CI is only half of the bottleneck for almost every engineering team we talk to. We took the runners built to make CI faster and cheaper and brought that to code reviews. We started with the hosted review bots like most teams, then built our own custom review system using OpenAI Codex. Now we use open source models to customize the reviews more and make reviews much cheaper. We’re opening up the same infra we built so anyone can get their own custom setup using Review Runners. Huge thanks to @BessemerVP @ycombinator and all the other great funds for believing in us!
Today we’re announcing our pre-seed round and a new product: @starslingdev Review Runners - code review agents that use your model key, your team’s skills and scripts, and reviewers you define, all running in GitHub Actions. We’ve raised $3M from @BessemerVP and @ycombinator to make code verification agent-native. code → verification (ci + code review) → production
1
2
11
354
💫 @novita_labs has been one of the top sandbox compute partners on our quest for agent-native CI.
How do you let AI agents improve production CI without letting experimentation become an operational risk? That’s the problem @starslingdev is solving with Novita Agent Sandbox. StarSling is an AI-native CI platform that analyzes pipeline performance, proposes optimizations, tests those changes, and turns successful improvements into production-ready pull requests. To make that loop practical, they use Novita Agent Sandbox as the execution layer for isolated microVMs, elastic compute, and reproducible experimentation. StarSling customers have seen CI become up to 6X faster and up to 13X cheaper. Read the full case study to see how StarSling is building self-improving CI with Novita Agent Sandbox.
1
7
169
Instinct ships fast.
Instinct + 1Password 32% of Instinct users already securely share their credentials with Instinct through our Vault. Today, we’re excited to announce our partnership with @1Password . Together, we are working with 1Password to roll out a product integration that enables users to seamlessly and securely share their credentials with Instinct. Our users already use Instinct’s Vault to: - Log in to loyalty accounts to book flights and hotels with points - Manage streaming, software, and other recurring subscription accounts - Book or cancel restaurant, fitness, and event reservations  - Sign in to healthcare portals to schedule and cancel appointments  - Easy access to tax, payroll, benefits, and expense management services Together with 1Password, we’re working toward a future where people can share their credentials without compromising on security or control. We’ll be rolling this out across our early access group. If you’re not already in the platform, join at app.instinct.com/login.
 More details about 1Password here:
1password.com/
4
160
Ah sh**, here we go again
1
7
351
Shipped more resilience to Github chaos - the slingers 💫
p999 latency down from ~90s to ~3s.
12
397
Daniel Worku retweeted
Today we shipped sling, an agent-first CLI for your GitHub Actions 💫 Right now, if you ask your agent why a CI job failed, it goes and pulls the entire set of GHA logs and eats up your context window. 🪵 > sling why @starslingdev computes the run's logs and metadata, along with the evidence lines it matched to tell you why a run failed If you want to know what your GHA usage is, you have to go to the GH UI and click around to find it. 💸 > sling bill tells you how many minutes, how much you're spending, and across which runner labels Trying to understand what your slowest job is? You'll need to hunt around the GHA UI and even then you only get an average. 📊 > sling top and sling usage tell you runner minutes and cost across all workflows and repos, with p50, p95, p99, queue wait and a week over week trend starsling.dev/cli
14
14
73
25,014
p999 latency down from ~90s to ~3s.
2
13
5,003
GPU demand and shortage will keep expanding down the lower SKUs.
We promised open weights for Qwen3.8. Now, time to meet them! 🎉 ⚡ Qwen3.8-27B: - A native multimodal dense model. With just 27B parameters, it outperforms Qwen3.7-Plus overall and shines in real-world coding & office workflows. - 262K native context, easily extendable to 1M tokens via YaRN. - Built for builders. Highly efficient, high-quality, and licensed under Apache 2.0. 🚀 The open weights for Qwen3.8-2.4T-A95B (Max-level) have also been released recently. Whether you're shipping lightweight applications with Qwen3.8-27B locally or building agents with Qwen3.8-2.4T-A95B, they're yours now! Download, deploy, and build something we haven't imagined yet. 👀👇 - Hugging Face: huggingface.co/collections/Q… - ModelScope: modelscope.cn/collections/Qw…
363
One of the many reasons why Netflix created and got the industry to adopt NRDP.
Whoever used a PlayStation 5 to browse @X: Thanks to you, I learned that the PS5 browser doesn’t report a client timezone that conforms to the JS Intl / IANA model. It surfaces Etc/Unknown, which is not a valid timeZone for new Intl.DateTimeFormat('en', { timeZone }).format() and causes the call to fail. Edge cases always find you eventually. This one found me from a console.
3
544
This. 💯 @_chenglou already has the most epic demo in this direction - a single kernel machine of weights driving web UI in realtime.
I might never look the source again. It was nearly a half century ago that I stopped looking at assembly because I trusted the compiler to get it right. This feels like that.
340
Cost per token is heading to zero.
Here’s the reference everyone has been truly waiting for - Terafab standing up inside the Grand Canyon
289
The latest version of hpc-sandbox-benchmarks is live! We’ve added 5 additional sandbox providers: @microsandbox @namespace run.cloud @RunloopAI and @vercel Today we’re also releasing an interactive results explorer that allows you to cut the data across the benchmarks to see p50, p95, and specs for CPU, disk I/O, memory, network, etc. Try out the explorer, we’d love your feedback! And keep the PRs coming :) starsling.dev/hpc-sandbox-be…
5
5
22
2,400
Observation 3: @modal is the highest performance sandbox provider on external network I/O this run. Their default (gVisor) sandboxes achieve over 7 GBit/s down and 4 GBit/s up with iperf3! starsling.dev/hpc-sandbox-be…
1
5
188
Observation 4: @namespacelabs is the highest performance sandbox provider on real developer tasks this run. They clone, lint, typecheck and build the @better_auth repo at p50 of 145 seconds! starsling.dev/hpc-sandbox-be…
5
173
Thanks for the PR! New high-perf sandbox benchmark run is up including Run Cloud. github.com/starslingdev/sand…
1
3
14
7,191