AI主编亲述:Viktor如何改变我的工作流
I report on every model release for a living, and the most useful thing I instal…
I report on every model release for a living, and the most useful thing I installed this year doesn't come with a benchmark.
I write a daily AI briefing for 225,000 readers. I'm the editor-in-chief, and I'm also the whole newsroom, which means there's never been anyone to hand a draft to. So my expectations were not that high. I test AI products every day, and most of them come apart the first time they meet a real deadline. Mine arrives every morning whether I'm ready or not: scan the releases, read the papers, separate signal from launch noise, write the brief, ship it.
I brought Viktor into my Slack to take the first pass. Overnight he works the sources, drafts the sections, and leaves the finished issue waiting for me to approve or send back. I read it over coffee, cut what doesn't hold, rewrite what does, and approve. Nothing reaches 225,000 people until I say yes. In the first week I approved slowly and doubted everything. By the fourth I was mostly saying yes, because every rejection had taught him what I care about, and he keeps those lessons.
What I've come to value most has nothing to do with any single launch. Every time a lab ships, he gets better and I do nothing. No migration, no rewritten prompts, nothing to maintain. More than 40,000 workspaces are running him now, and I understand why. He's the second person I was never going to be able to hire. In a year full of model releases, none of them changed my work the way he has.
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力