YC Startup Tracker

Both developments point to a growing need for infrastructure around agent-driven software work: Blacksmith is scaling continuous-integration capacity as coding activity rises, while Y Combinator’s QM organizes shared agent workflows across distributed teams.

The tension to watch is between speed and control, as larger-scale automation brings both failure-diagnosis tools and more explicit permissions, credential separation, and command policies.

Tracking: Y Combinator

Geography: Mountain View, Silicon Valley, United States

1. Blacksmith Raises $45 Million to Expand Continuous Integration Infrastructure

Blacksmith, a San Francisco cloud infrastructure company, raised $45 million in Series B financing at a $550 million valuation. Peak XV Partners led the round, while existing backers Y Combinator and GV participated.

The company provides dedicated infrastructure for GitHub Actions, helping developers build and test code continuously rather than relying entirely on GitHub-hosted runners.

Blacksmith says more than 6,000 companies use its platform, up from about 800 when it announced its Series A last September, and continuous integration jobs have grown 5% to 10% weekly since the start of 2026.

It plans to use most of the capital to expand from hundreds of thousands of CPU cores by roughly tenfold, while developing Codesmith, an agent that can diagnose CI failures and make fixes.

The fundraising positions validation infrastructure as a response to coding agents producing more pull requests, though the article gives no revenue or profitability figures.

Key facts:

- Blacksmith raised $45 million in Series B financing at a $550 million valuation.

- Peak XV Partners led; Y Combinator and GV participated as existing investors.

- More than 6,000 companies use Blacksmith, versus about 800 at its Series A announcement.

- Continuous integration jobs grew 5% to 10% week over week since early 2026.

- Blacksmith expects computing capacity to increase roughly tenfold in coming months.

Why it matters: Blacksmith is betting that software written with coding agents will make testing and validation infrastructure more important.

If its reported workload growth continues, dedicated computing and caching for continuous integration could become a larger software infrastructure category, benefiting Blacksmith and teams seeking alternatives to GitHub-hosted runners.

The funding also marks a shift beyond running CI workloads: Codesmith is designed to participate in diagnosing and fixing failures.

That could give Blacksmith a broader role in development workflows, but the company must scale its computing footprint while competing with Buildkite, CircleCI, and other CI platforms.

Citybiz

2. Y Combinator Open-Sources QM for Collaborative Startup Workflows

Y Combinator has introduced QM, an open-source “multiplayer agent harness” for coordinating shared work across distributed teams, according to Geeky Gadgets.

The cloud-based platform combines isolated workspaces, Slack integration, shared memory, and a unified way to manage tools, files, and processes. QM uses PostgreSQL to store session history and support agent loops, allowing teams to revisit earlier actions.

It also separates shared and private spaces for tools and credentials and gives administrators three agent-action settings—strict, auto, and dangerous—alongside command policies.

The article says QM remains early-stage, with interface, documentation, and administrative complexity still limiting broader adoption.

Key facts:

- Y Combinator introduced QM as an open-source, cloud-based agent collaboration platform.

- QM integrates isolated workspaces, Slack, shared memory, tools, files, and processes.

- PostgreSQL stores session history and supports QM’s agent loops.

- Administrators can choose strict, auto, or dangerous agent-action controls.

- QM’s reported challenges include interface design, documentation, and administrative complexity.

Why it matters: QM gives startups a customizable way to centralize work among distributed teams while preserving shared workflow history and controlling what AI agents can do.

Its open-source model may also let organizations adapt the platform and contribute to its development. The main constraint is usability: teams may face friction from the interface and administrative demands, especially while QM is still early-stage.

Better documentation, simpler administration, and interface improvements will determine whether it moves beyond technically capable early adopters.

Geeky Gadgets