30 秒でわかる
In this February 2026 short, Matthew Berman shows an OpenClaw assistant researching a proposed video and adding a task to Asana. The accompanying prompt describes a broader workflow for research, duplicate checks and concise production briefs; its similarity threshold is a configuration choice, not a guarantee of accuracy.
- 1Start with an ideaBerman asks his assistant to find sources and visual material for a proposed video.
- 2Create a production taskThe assistant adds a card to an Asana video pipeline with researched material.
- 3Research and deduplicateThe description supplies a longer prompt for searching sources, comparing past ideas and creating a brief.
ポイント
- A chat message can trigger research for articles, videos, posts and visual material related to an idea.
- The example creates an Asana task containing research links and a production outline.
- The supplied prompt proposes similarity-based duplicate checks, but the short does not establish their accuracy.
チャプターごとに整えた全文を読む
From an idea to a task
Berman demonstrates research from a chat request and shows the resulting Asana task. He shares a broader prompt for a content pipeline.
My team shares articles that might make interesting videos. In this OpenClaw example, I tell my assistant that an article is a video idea and ask it to find relevant articles, videos, posts and graphs I could show alongside it.
When the research finishes, the assistant creates a card in my Asana video pipeline. I open the card to show source material and a video outline assembled from the article; the short shows a result, not a systematic assessment of source quality.
The video description provides a broader content-idea prompt: research a topic, compare it with past ideas using semantic and keyword similarity, and create a concise task if it passes a duplicate check. Its proposed 40% cutoff is a chosen rule and can produce false matches or miss duplicates, so the output still needs review.
同じチャンネルの記事
25:50EN
32:20EN
0:22EN
45:08EN
0:32EN
12:14EN