Search for a command to run...
We're building the infrastructure layer that makes building and streaming non-linear agents equipped with tools as easy as writing one function call.
We are excited to announce that Dedalus Labs has raised $11 million in seed funding to help developers build complex agents across any model and any tool in 5 lines of code. Kindred Ventures and Saga Ventures co-led the round, with participation from E14 Fund, Emergence Capital, Sunshine Lake, Transpose Platform, Liquid 2 Ventures, FPV Ventures, Twenty Two Ventures, Telescope Foundation, Spot VC, Operator Partners, and Y Combinator.
We are fortunate to be backed by a group of angels behind some of the most important developer platforms and foundational AI research, including Thomas Wolf (Hugging Face Co-Founder/CSO), Cal Henderson (Slack Co-Founder/CTO), Ant Wilson (Supabase Co-Founder/CTO), Thomas Dohmke (former GitHub CEO), Tri Dao (FlashAttention; Together AI Chief Scientist), and Elad Hazan (AdaGrad; Google DeepMind & Princeton University).
We believe agentic workflows are really just agents—orchestrations of models and tools. An agent equipped with the right tools and guardrails should be able to execute any workflow without engineers hard-coding every edge case.
Your job as a developer should be simple: define the prompt, select the right tools and guardrails, and let your agent run.
Traditional workflows are not flexible, composable, or scalable. They are either linear drag-and-drop interfaces that are clearly not built for developers, or they lock your stack into one model provider. Worse, few are production-ready.
The right abstraction isn’t another visual editor or spaghetti framework, it's a developer-native infrastructure layer that makes building and streaming non-linear agents equipped with tools as easy as writing one function call.
My co-founder Windsor and I have always had high standards for what a good developer experience feels like. We built Dedalus because we were tired of settling for tools that didn't match how agents interact with the real-world.
Our first version abstracted away all the friction of deploying MCP servers. What used to take days of Docker files and YAMLs now takes three clicks.
But then, we quickly realized that even with hosted tools out-of-the-box, developers still struggled to build robust agents—parsing, recursion, multi-tool orchestration—no framework actually connected everything together.
So, we built the Dedalus SDK. Now, with 5 lines of code, your agents can execute vendor-agnostic model handoffs, chain both local tools and hosted MCPs, and stream across any provider.
We were tired of doing things the hard way.
Now, you don't have to.
Before, building an agent (model equipped with tools) meant spending days stitching together bespoke APIs. Then, Anthropic introduced Model Context Protocol (MCP).
MCP is an open protocol that lets AI models talk to external tools in a standardized way. Think of it as an API that AI models know exactly how to call: reliable, predictable, and language-agnostic.
As more companies adopt MCP, exposing your service as an MCP server means you're not just serving humans—you are also serving agents. In a world where agents are users, this is the next big distribution channel. We're building the infrastructure to power that shift.
Open-source is core to our mission, and we will continue to give back to the developer community. Next, we're open sourcing the first industry-grade MCP Authorization Server, a pioneering and production-ready solution that allows MCP Resource Servers to delegate token issuance, discovery, and security policies to a dedicated Authorization Server that tracks the MCP specs with high fidelity.
We will continue to build community around tool use and post-training, and we’re relentless about setting new standards grounded in how the agentic ecosystem ought to evolve.
We are also expanding beyond infrastructure. Dedalus isn’t just a developer tool; we are becoming the backbone for how companies build and distribute. Not just building agents, but shipping, sharing, and growing them. We are here to help you win.
Finally, this December we are launching Break In, a 3-week hacker house in partnership with The Residency, designed to help high-signal founders break into Silicon Valley. More on that soon.
At Dedalus Labs, we’re building the next generation of intelligent agents and systems that think, plan, and act with purpose. If that excites you, start exploring what we’re building today.
If you care about doing things the right way, join us! We look for people who care deeply about doing things the right way. Skill matters, but character matters more. If you’re sharp, humble, and driven to build something that lasts, we’d love to hear from you.