AI · ផ្លូវការ
Superhuman Go Shares Lessons From Routing Requests Across 100-Plus AI Agents
Superhuman blog ·

មើលត្រួសៗ
An applied research scientist at Superhuman published research on how the Superhuman Go assistant picks the right tool from more than 100 agents, a step known as routing. The post looks at why wrong picks become more likely as tools pile up and names embedded shortlisting, which narrows the candidate tools first, as the most promising fix.
ចំណុចសំខាន់ៗ
- Go has over 100 agents and 800 tools, ranging from single-purpose ones like Weather to Gmail with more than 15 actions.
- Earlier research found AI tool-calling performance fell by 7% to 85% as the number of tools grew from 49 to 741.
- Go must pick the right agent and then the right tool inside it, and look-alike tools such as Google Docs and Coda Docs make mistakes easy.
- The team built an evaluation framework to measure routing with three models: GPT-5.1, GPT-5.4, and Claude Sonnet 4.5.
ហេតុអ្វីវាសំខាន់
At Superhuman's scale even a 1% failure rate can affect hundreds of thousands of customers, so choosing the right tool matters more as assistants add features.
ប្រភពដើម
Agent Orchestration at Scale: Lessons From Superhuman Go
Superhuman blog · Oct 15, 23:26 KST
Read on Superhuman blog ↗AI summarized the original article. Check the original for accuracy. Copyright belongs to Superhuman blog.
បន្ថែមពីវិស័យនេះ
- Labour MP Joins Data Centre Committee While Opposing Norwich Project
A UK Labour MP opposing a large data centre near Norwich has joined a new government committee that will help shape national rules for these facilities. He says data centres are needed, but wants the Norwich Apex site stopped and built differently.
BBC News · Nov 21, 15:37 KST - OpenAI Opens ChatGPT Space to Plus Users for Pages, Slides, and Team Sharing
OpenAI opened ChatGPT Space to Plus users on the 20th, local time. Space works as a new home for pages, slides, images, files, and uploads. Users can write pages with ChatGPT, add interactive charts, turn ideas into slides, and invite others to edit together.
AI 매터스 · Nov 21, 11:26 KST - Anthropic Researcher Claims Claude Helped Prove Riemann Hypothesis, Experts Await Review
Wes Roth covers a claim by Anthropic researcher Levent Alpöge that he proved the Riemann Hypothesis with help from the AI model Claude. He looks at the huge Lean formal proof, reactions from mathematicians and AI safety researchers, and what this could mean for AI and math.
Wes Roth · Nov 21, 11:24 KST