Fall 2026
Cloud platform for agentic AI development
No salary listed
San Francisco, CA, USA
In Person
Five days per week on-site in San Francisco. Relocation support may be available.
See people who can refer or advise you
Dedalus Labs provides a cloud platform that acts as a drop-in MCP gateway, connecting any LLM to any MCP server with hosting, scaling, and a single API. Its OpenAI-compatible SDK lets developers orchestrate agentic AI workflows by routing between LLM providers and tools from a hosted MCP marketplace, without needing Docker or YAML configuration. The platform supports universal model access across models like GPT-4, Claude, and Gemini, and hybrid tooling that combines Python functions with pre-built MCP tools, along with serverless scaling and hot-reloading for production-ready deployments. It plans a marketplace for developers to monetize agents or servers with an 80% creator share and instant payouts, aiming to speed up AI development from weeks to minutes and simplify deployment, scaling, and monetization of agent-based AI workflows.
Company Size
11-50
Company Stage
Seed
Total Funding
$11.1M
Headquarters
San Francisco, California
Founded
2025
See people who can refer or advise you
Help us improve and share your feedback! Did you find this helpful?
Remote Work Options
Relocation Assistance
Company Equity
Meal Benefits
Introducing Dedalus Auth. Dedalus Auth (DAuth) is a multi-tenant authentication layer for MCP servers and marketplaces that protects user credentials and closes security gaps without requiring developers to become security experts. Introducing Dedalus Auth (DAuth). Today Dedalus Labs is excited to launch Dedalus Auth (DAuth), a new way to secure Model Context Protocol (MCP) servers and marketplaces. Its authentication architecture protects user credentials and solves long-standing security gaps in the MCP landscape. DAuth is designed specifically for MCP: it's multi-tenant, faithful to the MCP specs, and ensures that no one can see your secrets, not even Dedalus Labs. Existing auth implementations are underdeveloped and difficult for smaller developers to implement. API keys and raw secrets are still being passed around with limited protection, making them vulnerable to malicious servers that hijack user credentials. To build MCPs with production-grade security, developers need to spend weeks becoming security experts to build authentication and security infrastructure. Most don't bother. But Dedalus Labs believe you shouldn't have to compromise on security. That's why Dedalus Labs developed DAuth, so MCP creators don't have to reinvent auth infra, and users never have to hand secrets to strangers. The best thing about DAuth is that it is built in to its SDK, takes minutes to integrate, and doesn't require developers to be security experts. Security is the bottleneck for MCP adoption. The biggest barrier to deploying AI agents in production is trust. Recent high-profile security incidents have shown that MCP can be a vulnerable vector of attack for malicious actors. If Dedalus Labs is going to give agents the tools they need to be useful, their permissions must be handled accurately and securely. When Dedalus Labs started building its MCP marketplace, Dedalus Labs were disappointed by the state of MCP auth. MCP standardized around OAuth, but it doesn't support dynamically passing API keys, forcing existing MCP marketplaces to be OAuth-centric and single-tenant. In addition, the MCP specs don't specify how non-OAuth credentials should be handled. This is a problem because most MCP servers are non-OAuth, as many of the tools developers use are not authenticated by OAuth, like a search API or database credentials. For developers using those tools, MCP security becomes a chore. To do things the "right way," you need to create your own OAuth app for every tool. The easier route, storing raw credentials with the host, turns marketplaces into a single vector of attack: one malicious actor can collect thousands of secrets. No existing MCP authentication solution serves its users security needs. No one else seems to be working on one that will. So Dedalus Labs built the missing baseline: secure, multi-tenant auth that doesn't depend on every tool being OAuth. At the core of its authentication layer is a multi-tenant infrastructure. Single-tenant authentication means that every deployed MCP server needs to deploy its own authentication layer. Multi-tenant authentication means servers using its SDK can authenticate with DAuth, serving as a single authentication layer for every MCP server Dedalus Labs host. Its unique architecture is zero-trust and host-blind, which means Dedalus never sees raw API keys or access tokens. Here's what that looks like in practice: * The Dedalus SDK performs client-side encryption on your credentials, protecting secrets before they leave your device. * Whenever the SDK needs permission to do something, Dedalus Labs initiate a secure exchange, calling its open-source auth server to validate access. * When a user request reaches your MCP server, the server acts as a standard OAuth 2.1 Resource Server. Access is verified without ever touching the credential. * Once the MCP server validates a request, Dedalus Labs send it to The Enclave, a network-less hardware-secured enclave written in Rust. * Within this network-less Enclave, your credentials are checked for milliseconds before being zeroed from memory. * Then, the request is encrypted before being sent downstream, without ever revealing the secret to Dedalus or your MCP server. The Enclave is a digital Fort Knox, a network-less fortress built in Rust that neither remote hackers or Dedalus can see your secrets. Best of all, this works across all auth cases. This lets Dedalus Labs offer much stronger security guarantees than any existing marketplace or solution. With DAuth, Dedalus Labs is proud to offer the only secure and multi-tenant marketplace to date. If you're interested in learning more about its system, check out its technical blog post explaining how Dedalus Labs built it. Now that Dedalus Labs can trust agents, what's next? The best part about DAuth is that you no longer need to be a security expert to build production-grade MCP-powered agents. DAuth is built into the Dedalus SDK. Rather than implementing OAuth flows that change with each tool, developers define an Intent (e.g., slack_read) that ensures permissions can't be hijacked by malicious actors. This intent-based system can be implemented in a few lines of code, removing the burden of security from the developer. For users, this means they can trust MCPs made by anyone using the DAuth framework the same way they would trust an MCP from a big name like Google. DAuth opens up a new world of possibilities, where trust is a default on its platform. Dedalus Labs believe MCP won't matter in production until auth stops being optional and secrets are no longer out in the open. Security is the key to creating a robust, community-driven agent economy, where trust is essential to both creators and end users. Dedalus Labs is rolling out updated docs and example use cases to show you how to use DAuth to build useful agents in the real world. If MCP is going to power real-world agents, security can't be optional. When you build or run on Dedalus, you're opting into a standard you can trust. If you want to try auth today: * Build with Dedalus Labs Authentication Layer for FREE by deploying an MCP with its SDK. Free credits available until Jan. 27. * More use cases coming soon.
Dedalus Labs raised $11M in seed funding to help developers build complex agents across any model and tool in 5 lines of code, led by Kindred Ventures and Saga Ventures.
Dedalus Labs secures $11 million in seed round, launches build & deploy complex agents in 5 lines of code. ["Build and deploy complex AI agents with any model and any tool in just 5 lines of code. No Docker files. No yamls. No wasted weeks."]. The founders story: two princeton CS students with high standards. Meet Cathy (ex-Voyage AI / Salesforce) and Windsor (ex-DeepMind / Sentient AGI). They first ran into this problem while working with MCP during Windsor's time at Sentient AGI. Every solution felt wrong: drag-and-drop GUIs, brittle configs, frameworks that broke under real workloads. Nothing met their standards for what good developer infrastructure should be. So the founders built what they always wanted - a developer-first platform that makes agentic workflows composable, scalable, and effortless to ship. If no one else is going to build it right, Dedalus Labs will. The problem: building agents is hard. Building tool-using agents | calling v1/chat/completions. * Running their MCP servers locally. * If deploying remotely, wiring up AWS and config hell. * Juggling model quirks, streaming issues, and auth across GPT-4o, Claude 3.5, Sonnet, Flash, etc. ( | API keys). * Rebuilding everything just to add a new tool. The Dedalus Labs founders were tired of writing hundreds of lines of code and battling Docker files and infra - all for a spaghetti agent that breaks under stress. What Dedalus Labs offers: complex agents in 5 lines of code. Dedalus invented the simplest way to build and deploy agents - a drop-in infrastructure layer that unifies models, tools, and orchestration. * Vendor-agnostic model handoffs. * Chaining both local tools and hosted MCPs. * Real-time streaming across any provider. And their managed infra lets you deploy an MCP server in 3 clicks, so anyone can equip their agent with tools. No Docker files, no setup. Why MCP: tools made for agents. MCP (Model Context Protocol) is becoming the standard for how models talk to tools. Think of it as an API that AI models already know how to call - reliable, predictable, and language-agnostic. As more companies adopt MCP, exposing your product as an MCP server means you're not just serving humans anymore - you're serving agents. At Dedalus Labs, they're building the infrastructure that powers that shift. Learn more. [ | visit] [www.dedaluslabs.ai] [to learn more.].
Dedalus Labs has secured $11 million in seed funding to aid developers in creating complex AI agents. The funding round was co-led by Kindred Ventures and Saga Ventures, with additional 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.
Y Combinator joins $11m round in Dedalus Labs to develop AI agents. Dedalus Labs has raised US$11 million in seed funding to help developers build complex AI agents. The round was co-led by Kindred Ventures and Saga Ventures, 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. Notable individual investors include Thomas Wolf, co-founder and CSO of Hugging Face; Cal Henderson, co-founder and CTO of Slack; Ant Wilson, co-founder and CTO of Supabase; Thomas Dohmke, former GitHub CEO; Tri Dao, FlashAttention creator and Together AI chief scientist; and Elad Hazan, AdaGrad creator and researcher at Google DeepMind and Princeton University. Dedalus Labs develops infrastructure and software development kits that help engineers build and deploy AI-powered agents using multiple tools and models. The company plans to open source its MCP Authorization Server, which manages security and token policies for AI models using the Model Context Protocol. Food for thought. Model Context Protocol (MCP) adoption remains unclear despite rising interest. * Anthropic introduced Model Context Protocol (MCP) in November 2024 1. OpenAI uses it 2. Google DeepMind support is coming 2. Zed, Replit, and Sourcegraph back it 1. Metrics are missing. * The protocol lacks standardized authentication (identity verification) and authorization (access control) mechanisms 3. SaaS multi-tenancy hosting remains hard 3, and security risks like prompt injection plus tool permission gaps can enable data exfiltration 2. * Most MCP servers install locally (on a developer's machine) 4. AWS and Microsoft Azure are working to make MCP portable and cloud-compatible 4. The ecosystem remains experimental, which falls short of Dedalus's "5 lines of code" pitch at scale. SaaS companies can reach agents by building MCP servers now. * For business-to-business (B2B) software vendors, turning services into MCP servers gives AI agents and developers a direct path without middlemen platforms, like how REST APIs opened programmatic access in the 2010s 3. * Mintlify, Stainless, and Speakeasy offer tools (developer tools that turn API documentation into working servers) to produce MCP servers from docs 3. Cloudflare (a cloud and edge network provider) and Smithery (a platform for publishing and hosting MCP tools) provide hosting that eases setup 3. * Discovery options include mcpt (a command-line discovery tool for MCP tools), Smithery (a public registry), and OpenTools marketplaces (directories for AI tools) 3. The MCP Servers directory (a listing of MCP servers) adds reach 5. Early entries gain visibility, but results hinge on developer and enterprise uptake that remains unproven. How would you feel if you could no longer use Tech in Asia?