Back to directory
Enterprise infrastructure · Identity & auth

Didit

Infrastructure for identity and fraud.

CEO @getdidit (YC W26)
San Francisco, CA6.8K followers
TLVC Rating
Hook
Editing / Creativity
Copy
Sentiment of launch
Distribution strategy
Community Rating
No ratings yet
Your rating
Sign in to rate this launch.

About

Didit is building a single API for identity verification, KYC, AML, and fraud prevention aimed at developers and, increasingly, AI agents that need to prove who and what they are transacting with. The company was founded in 2023 in Barcelona by twin brothers Alberto and Alejandro Rosas, with headquarters now in San Francisco, and went through Y Combinator W26. The product bundles document verification, liveness, face match, and AML screening behind one integration, with coverage across 220 plus countries and pricing positioned well below legacy IDV vendors. This launch marks a $7.5M seed round backed by Y Combinator, Orange Collective, Rebel Fund, and Pioneer Fund, alongside angels. The thesis behind the raise is that generative AI has made synthetic identities, fake businesses, and fraudulent media cheap to produce at scale, and that nearly every consumer and B2B application will soon need a programmable trust layer to defend against it. Didit's pitch to developers is integration in minutes rather than the multi week compliance projects typically required to ship KYC. Alberto Rosas serves as CEO and Alejandro Rosas as CTO , and the company reports selling tools that businesses use to verify people, companies, digital wallets and transactions to a customer base that already spans thousands of organizations globally. For founders and operators evaluating identity infrastructure, Didit is worth a look if you want one endpoint to cover verification, compliance, and fraud signals rather than stitching together separate vendors.
Tags
<500KSeedExplainerB2BGlobalVertical AIFunding announcementFounder-led
Comments (14)
Sign in to join the discussion.
Priya Raghavan6d ago

Identity infra is having its moment and Didit has been quietly heads-down for a while. Watching this category closely.

Tomek Wojcik6d ago

The tweet just cuts off mid-sentence on 'fake'. Genuinely cannot tell if that's a thread hook or a misfire but I respect the chaos either way.

Lena Boateng6d ago

ok wait, the framing of 'AI made fraud cheap so we sell the antidote' is the cleanest pitch arc I've read this week.

Femi Adekunle6d ago

If the end user has to install an app or scan a QR with three steps, you've already lost half the funnel. KYC UX is where these companies win or die.

Dipti Narang6d ago

Curious where the biometric templates live and which regions you support residency in. Identity infra without a clear GDPR story is a hard sell to anyone with a legal team.

Wei Pang6d ago

The real question is p99 latency on the verification call and what your liveness fallback looks like when the model's confidence tanks. Marketing site says nothing about either.

Mateo Salinas6d ago

The wordmark is fine but 'Didit' in lowercase next to that screenshot crop feels cramped. Give the logo some air, it earned it.

Ravi K.6d ago

We built something adjacent to this internally at a big co years ago and the hardest part was never the verification, it was the appeals workflow. Curious if they've thought about that.

carlitos6d ago

Infrastructure for identity and fraud sounds like you're selling to both sides of the trade. Bold go-to-market.

Brendan O.6d ago

Still don't understand why this can't just be Plaid plus a selfie. Someone explain it to me like I just woke up from a coma.

Yuki Tanabe6d ago

Tagline rewrite, no charge: 'Proof of human. At scale.' You can venmo me.

Noa Friedman6d ago

Wild how identity verification used to be a side project with Stripe Identity and a Zapier. Now it's a category with real infra companies and that rules.

Hossein A.6d ago

Every fraud ring is going to be agent-orchestrated within a year. The infra play is correct, the question is whether detection keeps up with generation.

Saoirse Doyle6d ago

Naive q: if AI can fake an ID this cheaply, what stops it from faking the liveness check too? Like, genuinely asking.