StarSling, a San Francisco startup promising to slash the costs and wait times that plague software teams every time they push code, recently closed a $3 million pre-seed round led by Bessemer Venture Partners. The funding drew a crowd of believers: Y Combinator, where the company went through an accelerator batch earlier this year, joined the round alongside Vermilion Cliffs Ventures, Precursor Ventures, Cervin Ventures, Outset Capital and Transpose Platform. The company declined to name a valuation.
The pitch is deceptively straightforward. StarSling sells what it calls "AI-native" runners for GitHub Actions—the automated workflows that test, build and deploy code—paired with agents that watch how those jobs execute, then open pull requests suggesting ways to speed things up or cut waste. The company claims its infrastructure can reduce CI pipeline costs by up to 13 times while running builds up to six times faster than what developers get from GitHub's standard setup.
It's a market crowded with alternatives—Namespace, RunsOn, WarpBuild, Blacksmith, Depot—but StarSling's founders believe the agents, not just faster compute, will set them apart. The bet attracted an unusual roster of angel investors: David Cramer, who co-founded error-tracking company Sentry; Zach Holman, GitHub's second engineering hire; Idris Mokhtarzada, co-founder and CTO of personal finance app Rocket Money; and Kulveer Taggar of Phosphor Capital. Engineering and product leads from OpenAI, Anthropic, ClickHouse, Google, Microsoft, Meta, Coinbase and Shopify also put money in, according to the company.
The team has the credentials to back the ambition. Daniel Worku, co-founder and CTO, spent years leading the dozen engineers who built Netflix Console, the internal developer portal used by over 3,000 engineers at the streaming giant. Before that, he worked on Facebook's AI Camera team. Yonas Beshawred, the CEO, founded StackShare, a developer community that grew to more than 1.5 million users before FOSSA acquired it last August. Ali Kayes, a founding software engineer, came from stints at Apple, Disney Streaming and Cisco.
"Daniel helped build the internal developer platform the 3,000+ engineers at Netflix relied on, and Yonas built StackShare into a community of more than a million developers," said Elliott Robinson, a partner at Bessemer Venture Partners, in a statement. "Their combined experience and understanding of developers at that scale is why we backed their vision to bring AI agents to DevOps infrastructure."
StarSling has processed over 2 million CI jobs since launching to general availability on June 29, the company said. The platform runs on current-generation AMD EPYC processors, which StarSling claims deliver roughly 30 percent better performance than GitHub's default runners. Pricing starts at $0.004 per minute for the smallest instance and climbs to $0.128 per minute for the largest; the standard 4-vCPU, 16-GB configuration costs $0.008 per minute. New accounts receive 2,000 free minutes in their first month.
More recently, the company added GPU runners with options ranging from RTX 4090 and 5090 cards to PRO 6000 models, with H100 support on the roadmap. Alongside the funding announcement on September 22, StarSling unveiled Review Runners—AI-powered code reviewers that operate inside GitHub Actions workflows. The feature lets teams configure reviewer "skills" and instructions in their repositories; code review runs before a human ever opens a pull request, using a read-only token and pulling reviewer guidelines from a trusted base branch. Customers bring their own model provider keys, keeping billing under their control.
The optimization agents work by monitoring build logs, execution patterns and resource usage across workflows. When they spot inefficiencies, they propose changes—tweaking parallelization, adjusting resource allocation, or reworking job dependencies. The company claims the automation frees engineering teams from constantly tinkering with CI configurations.
Early customers offer evidence the approach works, at least in some cases. Mastra, which builds an AI framework, cut its test suite from nearly 30 minutes to just over 5 minutes—a sixfold improvement—with queue times dropping eightfold, according to case studies published by StarSling. "At Mastra we move so fast that the bottleneck becomes reviews and CI," said Abhi Aiyer, the company's co-founder and CTO. "The runners are just handled, so nobody on my team thinks about CI infrastructure anymore."
Better Auth, a project from Vercel's team, shaved its end-to-end job time from 2 minutes 22 seconds down to 1 minute 4 seconds and saved roughly 20,000 CI minutes monthly, according to founder and CEO Bereket Engida. Partcl, a peer from Y Combinator's cohort, reported costs per run dropping by a factor of 13 and queue times shrinking sixteenfold. "Within a day of migrating to StarSling Runners, their agents opened up a PR that literally made our Rust CI tests 2x faster," wrote Vamshi Balanaga, Partcl's co-founder and CTO.
The broader market for GitHub Actions alternatives has grown more complex recently. BuildJet, once a competitor, shuttered its operations earlier this year, according to Latchkey, a developer-tools site that tracks CI infrastructure. Meanwhile, GitHub itself trimmed Actions pricing in a move that tightened margins for third-party providers. StarSling's founders are betting that AI-driven optimization—not just cheaper or faster machines—will carve out defensible ground in a space where raw compute alone may not be enough.
The company plans to spend the new funding on compute infrastructure and hiring additional founding engineers in the San Francisco Bay Area. The Review Runners feature remains in early access as the team gathers feedback and refines the agents' capabilities. Whether developers will trust AI to both run and review their code pipelines remains an open question, though the early customer testimonials suggest at least some teams are willing to try.
For now, StarSling is making the familiar startup wager: that automating away a painful, repetitive task—in this case, optimizing CI workflows—will resonate with enough engineering teams to build a business. The founders have done it before, in Beshawred's case, and helped build the tools that made it possible, in Worku's. The $3 million gives them runway to find out if they can do it again.
