Founder of @numbersprotocol, @dt42io
cyber
Joined October 2010
- Tweets276
- Following255
- Followers236
- Likes494
Why you should work with a seasoned tech founder on your project: they’ll fill whatever gaps come up, in any way necessary. The truth is, you’ll always run into surprises when building something.
ProofSnap extension - preserving digital provenance for screenshots. If you need to take screenshots as proof, this extension is for you.
We are eager to get your feedback. AI is amazing for helping people unlock creativity. Numbers and provenance help make this vision on track.
Human truth, machine proof. It's not a problem that humans or AI create content. The problem is that we should have the freedom to check and verify provenance easily.
techcrunch.com/2025/11/12/ja…
Proud that @GooglePlay WeArePlay recognizes @numbersprotocol. Thanks to the team, the community, and everyone believes in Numbers' vision.
Numbers has been selected by Google Play's global #WeArePlay program. A campaign that spotlights impactful app founders and their products.
Better AI begins with better data, and with our solution, everyday asset turned verifiable.
Well, I can hardly imagine anyone working on a scam project for more than four years. In any case, take this as proof that my team has worked very hard to make NUN survive and keep delivering in the critical crypto market for so long.
Hi Numerators, I understand that the chart's lagging behind got you frustrated. Crafting a legit Web3 project consisting of NUM (economic product) and protocol (functionality product) isn't a cakewalk. The only reason the team keeps dedicated to it is simple: every digital content must come with a verifiable proof of its provenance. It's our northstar. We also believe in the magic of every great community. Every successful project must be with a great community, because everything is about people in the end. You have your magic power to grow it together.
If you're an early-stage startup founder, this post is worth your time. The challenge for me is that getting the right people on board is incredibly hard.
New Era
1/n
I’m really excited to share that our @OpenAI reasoning system got a perfect score of 12/12 during the 2025 ICPC World Finals, the premier collegiate programming competition where top university teams from around the world solve complex algorithmic problems. This would have placed it first among all human participants. 🥇🥇
My GF shared a practical way of reading a (business) reference book with me: Ensure you apply the concepts you learned to life successfully, then move to the next book. Before that, stick with the current book.
Scammers pair cloned handles with video calls to lower your guard.
Speaking from my experience: the weak point is the third‑party call app people install from an invite link.
Get apps from official sites, verify handles, never share screen, use cold storage, and never sign blind.
I just got drained for $996,000 🧵
I’m honestly still processing that this happened and even writing it out feels completely unreal.
I just spent the last 24 hours looking into it, wiping out all my devices, and moving my files.
Part of me was considering whether I should even reveal this or not but let this be a good wake up call for everyone in this space regarding security, because if it can happen to me, with all my experience in this space, it can happen to anyone.
If my loss prevents this from happening to even one person, then it wasn’t all for nothing.
Not asking for pity, money, or anything. I know I made a mistake. Just want to help others.
Great tips. Here’s another approach:
When you need to modify or add a function, ask the AI to make only the minimum necessary changes to the code.
This helps keep the output stable.
Prompt:
“Given this output: [paste], write the shortest prompt that would produce an equivalent answer. Include steps, tone, and key constraints.”
Another prompting trick I’ve found consistently effective is reverse prompting.
The idea is simple: take a good output that’s close to what you want, feed it back into the LLM, and then ask the model to generate a prompt that would produce that same output.