部落格
誠實的 AI 代理筆記——寫給寧願動手、不願等的 builder。
Where a $0.04 model beats a frontier model: how Tulsk uses Jev
We measured every small decision our agents make, found where a typed judgment model wins, and shipped it in shadow first. What we kept, what we rejected, and the numbers behind both.
閱讀文章
Your agents don't need better models. They need a shared brain.
The multi-agent context problem is not a model problem. It is a shared-memory problem humans and agents both need to read, edit, and trust.
Tulsk is joining riggd.ai
Tulsk is becoming part of riggd.ai — a new company built to bring AI Harness to Asian SMEs. Your account, data, and pricing don't change. Here's what's actually moving and why.
Coding agent orchestration vs agentic workspace
Multica orchestrates coding CLIs. Tulsk is the workspace where mixed teams plan, delegate, and ship work end-to-end. The two products operate at different, often complementary layers.
What is an agentic workspace?
Why setting up AI agents shouldn't take longer than building your MVP — and the missing layer between "a smarter model" and "a real teammate."