SafetyKit Raises $27m to Make AI do Tech's Dirty Work

Patreon, Eventbrite, Upwork, Substack, and the world’s largest marketplaces and payments companies rely on SafetyKit to keep their platforms safe.

Today big tech and finance companies rely on armies of low-paid outsourced humans to protect their platforms from fraud, harassment, and crime. It’s super important work, but the jobs are difficult and sometimes even traumatizing — and it’s impossible for companies to keep up.

It might seem like human review is the gold standard for this work, but here’s the truth:

• Human accuracy for BPOs is typically around 80%. One in five decisions are wrong!
• AI agents are better than humans at tasks that require infinite patience. Sometimes finding issues is like looking for a needle in a haystack. But it only takes one illegal product or scammer to seriously hurt your platform, and you need to find it.
• Humans are inconsistent and bring their own values. For example, someone from a different cultural background reviewing LGBTQ content can have a hard time understanding the relevant norms.

If any human work should be replaced by AI, it’s this.

SafetyKit builds AI agents that check for every bad thing you might worry about, plus the things you don’t know about like new scams and sanctions. That means companies don’t need to rely on overworked humans to do their dirty work — they can let AI agents do it instead.

With the right guardrails and frameworks, AI agents already outperform human reviewers. Not because they’re more intelligent, but because they’re more consistent, more patient, and work 24/7.

SafetyKit’s AI agents can replace outsourcing firms like Genpact and Accenture for risk, compliance, and safety work. Our agents perform complex workflows that require in-depth investigations and subject-matter expertise, from finding the cracks in a fraudster’s story to reviewing cannabis products against regulations for 20 different jurisdictions. We scale elastically to handle millions of reviews across thousands of workflows and policy areas — that’s impossible to do with humans.

Huge companies already rely on SafetyKit in production — companies like Patreon, Eventbrite, Upwork, Substack, and Character.ai. We’re also working with some of the world’s largest marketplaces and payments networks, including one of the four card networks. They’re offloading work from humans and solving problems that were intractable without AI.

It’s working: Upwork measured us against humans, and found 80% accuracy with humans and 95% with SafetyKit.

The technology is here, but enterprises need SafetyKit’s help to rapidly deploy it.

We’re building SafetyKit because we’ve lived these problems firsthand.

My co-founder Steven and I first met in 2011 on OkCupid — we were the only two people at the time who mentioned Python programming and enterprise software in their dating bios (really!). We became friends and started our first company together in 2012, building automation tools for huge enterprises like Cisco, Roche, and DirecTV. 

Later I went to Stripe, where I built the risk platform that scaled Stripe from 20 underwriters to 2,000 outsourced human reviewers. We got early access to OpenAI models at Stripe in June 2020, and the first person I called was Steven — it was clear that AI was going to transform back of house business processes for large enterprises (this is seriously what the two of us like to talk about).

I thought that transformation would happen overnight, but it turns out even when the technology exists, it requires careful training, monitoring, and internal knowledge-building to make it work. That’s what we do at SafetyKit.

We're raising new funding to grow our team and transform this industry.

We’ve raised $27M in funding from Ribbit, First Round Capital, Y Combinator, and notable angel investors Claire Hughes Johnson (former COO of Stripe), James Dyett (Head of Enterprise Sales at OpenAI), and Jon Zieger (former General Counsel at Stripe) to transform this industry as quickly as possible. Our mission is simple: free companies from manual toil by deploying AI that lets them focus on their business instead of risk operations.

If this sounds interesting to you, we’re growing the SafetyKit team and hiring engineers, designers, and risk experts in San Francisco. Check out safetykit.com/jobs for open roles. If you want to learn more about what SafetyKit can do for your company, email us at founders@safetykit.com — let’s talk.