every experiment taught us something. every node that stayed online, every bug report, every piece of feedback, and every late night spent testing alongside our community helped shape what zkagi is becoming.
the sandbox was never the destination - it was the beginning. we launched ideas before they were perfect because the only way to build infrastructure that lasts is to test it in the real world, learn from every failure, and improve with every iteration. that journey gave us invaluable lessons about performance, privacy, security, and what enterprises actually need to trust ai at scale.
today, we're taking those lessons into the next chapter. @zk_agi 2.0 is about turning experimentation into production-ready infrastructure without losing the values that brought us here. privacy, verifiability, and community remain at the core of everything we build.
to everyone who believed early, ran nodes, tested our products, challenged our assumptions, and helped us improve: thank you. your contributions are part of the foundation we're building on, and this next phase exists because of you.
the experiment continues - but now it's being built to last.
Single Node & 2-Colocated nodes Sharded have good performance stats with privacy
Still improving performance of Sharded Inference across 3 nodes in different continents, privacy holds
Read more: https://t.co/C6BX6EyFsi
Very proud day! Something that has kept me and Nikhil up for many nights over the last 2 years is how to bring mathematically verifiable privacy in AI (specially more complex models).
Early in our journey in 2024 we even attempted a Zkml circuit based approach only to find that it took 8 hours to return a simple output, next we moved on to accepting that GPU TEEs with Zk verification (which is along the lines of industry standard for privacy AI) would be a good approach; and perhaps federated learning as another reasonable approach.
FHE for all its mathematical beauty, is very slow and impractical.
The answer was in splitting the linear algebra, MPC based computations and targeted homomorphic encryption at “entry” and “exit” points.
And we didn’t just want to solve private AI and unlock the world of AI on sensitive data for one model, we wanted it to be extensible to any open source model.
Presenting to you the Altaica Compiler by @zk_agi
We decided to expand the gospel of @zk_agi across Nigeria, we dey Ogun state dey share the zk movement 😎
Open snooker championship 💪
Everyone is welcome to participate
Let's meet on 28/3/2026 by 5:00 pm UTC.
in this era of fast-moving markets and front-running bots, if you’re still trading manually, you’re leaving your edge on the table.
efficiency, speed and privacy matter more than ever.
i introduced you to PawPad, the first privacy-preserving AI trading app integrated with @zk_agi and Zcash.
it allows autonomous trading directly from your wallet without ever exposing your private keys or strategy.
in case you missed it, check the quoted post for a breakdown of what PawPad is and how you can create your keyless wallet to get started.
now, I’ll walk you through on how to configure your agent so it can start trading securely and intelligently.
after creating your wallet, follow these simple steps;
↳ click on Agent to begin.
↳ select your trading asset.
↳ choose maximum trade size
↳ choose your risk level: Conservative, Moderate, or Aggressive.
note that; Conservative limits exposure and focuses on safety, Moderate balances risk and growth, and Aggressive seeks high returns but increases volatility. Your choice directly affects how your agent executes trades.
all trades run inside Oasis TEE (Trusted Execution Environment). This ensures your private keys, strategy and trading logic remain fully confidential. even the agent itself cannot leak your data.
once you’ve configured asset, trade size, and risk, simply activate your agent.
from this point, your agent executes trades autonomously, according to your parameters, while keeping everything private and secure.
also, you can always edit your agent settings to adjust assets, risk, or trade sizes.
this flexibility allows you to adapt your strategy without compromising security.
PawPad combines automation, privacy and control, creating a trading experience where your strategy is protected, execution is precise and your keys remain fully yours.
enter Pawpad : https://t.co/WPUXBgvgXV
It was a pleasure for ZkAGI to be presented at Crypto Mountain Davos 🇨🇭!
A group of esteemed attendees from Tradfi, Crypto protocols and Quantum focused organizations attended the event
@AtenKrotos the founder of ZkAGI presented the technology, use cases
Healthcare meets Data Sovereignty with Zk Tech
Imagine controlling your health data like never before
Think of it like this;
Presently, patients have to choose between sharing their entire medical history or getting no help at all
But with this @zk_agi tech
▫️We don't need to provide a 50-page PDF of our personal information in order to show a vaccination or a "clean bill of health" for travel or work
▫️Hospitals can finally collaborate on rare disease research using this agent to query private datasets without ever actually "seeing" the patient's identity
Data sovereignty isn't just for finance; It's for our lives
Big ups to @zk_Terminal and @AtenKrotos 🫡
Merry Christmas 🎅 🎄
May this season bring warmth to your homes, inspiration to your minds, and strength to your journeys ahead.
Here’s to creating meaningful technology, supporting one another, and building a future where intelligence and humanity grow together.
From all of us at ZkAGI - core team and ambassadors, thank you for being part of this mission.
Have a magical Christmas and a wonderful New Year ahead. ❤️