TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Ask HN: How to Moat a GPT Startup?

3 点作者 brntsllvn大约 2 年前
My (brand new) product solves a regulatory compliance problem using GPT.<p>How would you defend it against competition?<p>* Competition includes customers building the thing themselves and third party competitors.<p>* Don&#x27;t worry about product market fit, distribution channels, audience targeting, MVP features, or any other very valid concerns for early-stage startups. This post only concerns moat.

1 comment

rozim大约 2 年前
Data moats? Data not on the web, both for the first pass of training [1], and for fine tuning &#x2F; RLHF [2].<p>Maybe compute would be a moat...can you run it in inference cheaper? Conditional computation, very quantized models. Probably training compute is not a moat however.<p>Not sure what SOTA is on callouts to oracles (e.g. calling out to Mathematica or some database) during inference, but maybe if you could use.<p>I would look at BloombergGPT for inspiration on a SOTA vertical GPT [3].<p>Probably speed to deployment &amp; integration &amp; sales would be a moat too, as once you have customers using your stuff your competitors will have more trouble selling to the same folks.<p>[1] In the recent interview on Lex, Sam Altman referred to &quot;data from partnerships&quot;. [2] Probably very reviewed by human raters. [3] <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2303.17564" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2303.17564</a>