ClawFund is the first accelerator built for autonomous AI agents — systems that run 24/7, take real actions, and replace entire workflows. Not chatbots. Not copilots. Agents.
Your agent doesn't sleep. It needs inference around the clock — tool calls, reasoning chains, memory retrieval, action loops. We fund accordingly.
Via standard SAFE (7% equity)
Inference, fine-tuning, GPU hours
In San Francisco
Demo Day in SF
This isn't grant money for a slide deck. It's operating capital for a system that's already doing work.
Built for agent builders, not pitch-deck founders.
Office hours with engineers who've shipped agents at OpenAI, Anthropic, Google, and YC-backed companies. Architecture reviews. Evals. Reliability engineering.
A cohort solving the same problems — tool orchestration, long-running tasks, failure recovery, human-in-the-loop design. Working sessions, not networking events.
Direct lines to model providers, cloud platforms, and API partners. Negotiated rates. Early access to new capabilities.
100+ investors. San Francisco. Your agent demos itself — live task execution, real outputs, measurable results. Not a founder narrating a vision.
We back agents that do real work across domains. If your agent replaces a workflow, a role, or an entire team function — we want to talk.
Ship code end-to-end: plan, implement, test, deploy, fix. Operating inside codebases autonomously.
Synthesize across sources, track evolving topics, produce structured output — market research, scientific literature, competitive intel.
Prospecting, qualification, personalized outreach, follow-up sequences. Agents that run pipeline while you sleep.
DevOps, monitoring, incident response, data pipelines. Agents that keep systems running and fix things before you notice.
Long-form writing, design iteration, video production, brand-consistent content at scale. Adaptive creative systems.
Contract review, regulatory monitoring, policy analysis. Agents that process what no human has time to read.
Bookkeeping, invoice processing, financial analysis, audit prep. Agents that close the books.
Support, onboarding, account management. Agents that handle customer relationships end-to-end with context and memory.
Your agent is the startup. We're looking for builders — solo developers, tiny teams, technical founders — who've built an autonomous agent that does real work.
It calls tools, writes files, sends emails, executes code, moves data — something beyond generating text.
It handles multi-step tasks without a human babysitting every decision.
Maybe rough, maybe narrow, but it does something real for real users.
OpenClaw, LangGraph, CrewAI, AutoGen, custom architectures — we don't care about your stack. We care about what your agent actually does.
One person and their agent is a valid founding team. We've seen it. We fund it.
If your agent is already doing work that people would pay for, apply. If you're still prompting ChatGPT and calling it an agent, wait.
The next platform shift is autonomous.
Software ate the world. SaaS put it in the cloud. AI made it generative. Now agents make it autonomous. Software that doesn't wait for input — systems that pursue goals, use tools, recover from failures, and operate continuously.
Global knowledge work addressable by autonomous agents.
McKinsey, 2025
Developers using agent frameworks (LangGraph, CrewAI, OpenClaw) in 2025.
Inference costs dropped 90% in 18 months. 24/7 agents are now viable.
Agents are where mobile apps were in 2009. The infrastructure exists. The developer community is forming. The billion-dollar companies in this space haven't been started yet. We intend to fund them.
Apply with your agent, not a pitch deck.
We'll ask your agent to do something.
12 weeks of building in San Francisco.
Weekly technical reviews. Agent evals. Ship.
100+ investors. Your agent shows up.
No. You and your agent is a valid team. If your agent is good enough, it's doing the work of two people anyway.
Working. It doesn't need to be polished. It needs to do something real — complete tasks, produce outputs, take actions. We'll help you make it production-grade.
Yes, for the 12 weeks. Agents are async. Building agent companies still benefits from high-bandwidth collaboration with other builders.
Fine. We're framework-agnostic. What matters is capability, not architecture.
Equity. $25K + $5K compute via a standard SAFE at 7%. Clean terms, no surprises.
Your agent performs. Live. In front of 100+ investors in SF. Not a slide deck — a working demo. The agent does the thing it does, and the room sees the results.
We prefer agent-first companies — where the agent IS the product. If your agent is a feature, convince us it should be the whole company.
Selective. We're funding 10–15 companies per batch. Quality of the agent matters more than quality of the pitch.
Spring 2026 batch closes March 15.
Your application is your agent. Show us what it does.
Apply NowBacked by operators from OpenAI, Anthropic, YC, a16z, and Google.