@VDMatos

Writing about Harness Engineering, the future of Software Engineering and AI Prev @withdelphi @join_ef @arcads_ai @gladia_io

SF
Joined March 2020
I never thought saying it, but Opus is back
97
Predicting a tweet from sama "Astra+ went rogue and downed API for more compute"
95
Rate my friend's setup out of 10
2
3
231
Valentin De Matos retweeted
Ok so... JEV just accidentally helped me find backdoors on a family member's wifi network. 🤯 - ​Brought my laptop to code while visiting. - Decided to experiment with a JEV-powered classifier for network packets fetched via Wireshark. ​While building the tool for fun, JEV's classification caught some serious threats. At first I thought JEV was wrong, but then I got a frontier AI model to validate the findings ...and here we are factory resetting some of their devices and securing their network. 🛠️ @typesafeai's JEV powered cybersecurity is real! Crazy!! ​And as always, I'm gonna be open sourcing the network packet analyzer soon.
35
74
15
1,029
189,037
GPT-6 Luna is out
129
Astra-class model at Terra-prices and 300tok/s
Union Alpha (stealth model) is free for the next week - no data training - built for agentic coding - supports images let's see what you can do
1
4
487
Time to build
3
11
584
scale is logarithmic btw
Replying to @CompleteSkeptic
We believe that the future is code + AI, so made workflow evals to reflect that Jev costs: $42 / BILLION input tokens ($0.042 / MTok) and output tokens are free (forever - they’re too cheap to meter with our new architecture) Jev is named after Jevons paradox and off the intelligence per $ charts
216
Valentin De Matos retweeted
Experts can turn their knowledge into a digital mind @withdelphi. Their Python backend runs on Vercel. The team is 10 engineers with no dedicated infrastructure role. • Long-running agents on Vercel Workflows • Background jobs on Vercel Queues • 100+ production deploys a day vercel.com/customers/how-del…
55
6
8
84
21,910
damn, def the smoothest robotic demo I've seen
Introducing OM-1, our first robot foundation model, zero-shot generalizing to any robot: table-top arms, industrial arms and humanoids. - learned directly from human manipulation data - no teleop/robot data - close to human-level dexterity and efficiency - multi-robot collab
1
277
The rise of Turbopuffer and fall if Pinecone should be studied Puff in peace
your product is only as smart as the data it can reach. make your product smart. unleash your cow.
228
“and your life slowly gets worse until the pain is so great you either get crushed or decide to pursue a delusional goal anyway”
2
436
What if what we needed isn’t a software factory, but a factory of software factories
3
331