@dttg retweeted
"Everything humans have built runs on trust, but that belief is breaking"
That's how our founders @albertorosasg and @_arosasg begin the letter on why we build Didit.
didit.me/manifesto/
@dttg retweeted
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
Readers added context they thought people might want to know
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-gri…
research.nvidia.com/labs/amri/proj…
tech-ish.com/2026/10/02/tav…
Kapwork Grow · See where AI buildout & infrastructure money is landing! kapwork.com/grow
@dttg retweeted
We merged nearly 2,300 PRs in the last 30 days alone at @didit, 99% automated. Been heads down this month powering a true Company Factory: a company that runs itself, tapping humans only when strictly necessary.
The volume of output is massive. There is nothing quite like waking up to 50+ new PRs ready to go. These are fully reviewed by other agents arguing until they reach a consensus (a literal PR AI war). By the time they hit my queue, they have AI approval, clean CI passes, and visual proofs attached.
I will be dropping a video soon breaking down exactly how this works under the hood. Stay tuned.
@dttg retweeted
If this is actually the new GLM 5.x Flash, z.ai just cracked RSI. Tang Hie was talking about self improvement and using synthetic data to train the next wave of models, and it looks like they finally figured it out.
We basically do not need human data for coding anymore. Agents are writing all the code these days anyway, and humans just steer the wheel. Once you build a system that knows exactly how to steer and get the exact results you want, you have completely solved synthetic data. That creates a direct loop for much better models. I believe this is what happened here
No way! @synthwavedd says that the secret Model "Ox Alpha" on OpenRouter is the upcoming GLM 5.3 Flash
A Flash model (!!) outperforming even the best models like Fable and GPT Sol?! If this turns out to be true, zAI's post training put western models at shame
Oh, and looks like Kimi k3.1 also incoming
What is happening right now. unbevlieable
@dttg retweeted
Unico's biometric database covers 96% of Brazil's economically active population. Four of the five largest private banks in the country run on it.
Today we partnered with Unico to bring that level of verification to Didit. Selfie plus CPF, no document, live in a Brazil workflow in minutes.
Unico IDCloud is now available on Didit for verifications in Brazil.
Selfie + CPF checked against a biometric database covering 96% of the economically active population. Answers in under 3 seconds, document capture only as a fallback.
didit.me/blog/unico-didit-pa…
@dttg retweeted
Unico IDCloud is now available on Didit for verifications in Brazil.
Selfie + CPF checked against a biometric database covering 96% of the economically active population. Answers in under 3 seconds, document capture only as a fallback.
didit.me/blog/unico-didit-pa…
@dttg retweeted
Fable will be back within days.
Letting frontier users migrate to Chinese models is strategically bad for both Anthropic and the US.
My guess: mandatory IDV, biometric auth, and much stronger anti-distillation protections. (That is why they PR the Alibaba distillation today)
The lazy question is: “Is Venezuela back?”
The better question is: “What gets built when an entire economy has to reconstruct its financial and digital infrastructure from scratch?”
We went to Caracas to find out.
blog.saasholic.com/p/venezue…
@dttg retweeted
Thinking about building your next company?
We want to help you make it happen. Before the deck, before the round, before the idea is even fully formed.
That's our new program: Second Time, First Check.
Apply: saasholic.com/second-timers
@dttg retweeted
Identical twins doing identity verification. Still cracks me up.
Congrats @didit on the fundraise!
We just raised $7.5M to build the infrastructure for identity and fraud.
The round is backed by @ycombinator, Orange Collective, Rebel Fund, Pioneer Fund, and other incredible VCs and angels.
AI is making fraud incredibly cheap to create: fake identities, fake businesses, fake transactions, all generated at internet scale. We believe every app will soon need trust infrastructure the same way apps today rely on payments or cloud hosting.
The future of the internet is verified. Didit (@didit) is building it.
@dttg retweeted
We just raised $7.5M to build the infrastructure for identity and fraud.
The round is backed by @ycombinator, Orange Collective, Rebel Fund, Pioneer Fund, and other incredible VCs and angels.
AI is making fraud incredibly cheap to create: fake identities, fake businesses, fake transactions, all generated at internet scale. We believe every app will soon need trust infrastructure the same way apps today rely on payments or cloud hosting.
The future of the internet is verified. Didit (@didit) is building it.
AI will 100x the identity market.
Every AI agent, synthetic user, fraud attempt, wallet, marketplace, fintech and regulated platform will need programmable identity infrastructure.
@Diditprotocol is building that layer.
Proud to back the team.
braziljournal.com/a-didit-qu…
@dttg retweeted
Check them out at: github.com/didit-protocol/di…
