Mehmet

Jerves is now in private beta. Jerves is a control plane for personal and company AI agents: it takes a request, picks the best specialist agent and model for it, and runs that agent with only the context it's allowed to see. For individuals, it keeps your memory and preferences so you don't have to repeat yourself across tasks. For enterprises, it turns workers' approved AI conversations into training data, kept separate per team, that builds a private model on the company's own server. Raw chat history never leaves that environment, and nothing is used until it has been scanned and approved. I’m opening access in waves over the coming days. Apply here, I'm letting people in over the next few days: jerves-wl-website.vercel.app…

Everybody talk about how AI improved but we have more important problem: how can we evaluate it. Check this

I shoot my shot for @ycombinator. Fingers crossed.🤞

app.jerves.online signup I will let you in ASAP

👀👀

I am rolling out access over the next few days. Until yours is ready, you can compete on the leaderboard in our Flappy Bird–style minigame while checking out what’s under the hood: Smart Routing: It sits between you and the models. Instead of relying on a single generic assistant, every prompt routes to the right specialist—whether that’s a coder, researcher, or critic. Privacy-First Memory: It retains your context while stripping out names, emails, and phone numbers before anything is stored. It also indexes and searches your local documents across English and Turkish seamlessly. Verified Execution: For development tasks, it writes the code, executes it inside a strictly air-gapped sandbox, and only marks the task done once all checks pass. Local First: Your data never leaves your devices.

New one just dropped do not forget to read!

Damn

Applied and excited

I wish I were studying robotics in 2000s

Artık @claudeai 'de Jev kullanabilirsiniz öncesinde bilmiyormuş

Figure just melted its old robot fleet Terminator-style after Arnold suggested it. NVIDIA added a hardware watchdog to stop AI agents from going rogue. Google shipped Gemini 4 Argon. Hardware is iterating so fast that last year’s robots are already scrap metal, while agent safety is becoming a chip-level problem. Which of these shifts matters more right now: speed of robotics, or keeping agents under control?