Pratham

I passed a long-form YouTube video into an API and it gave me back: • 10+ short clips • auto-animated captions • timestamped summaries • searchable moments • a full transcript with speaker labels All automatically. Normally, repurposing this kind of content is painful: Scrubbing timelines, finding highlights, clipping, writing captions, pulling timestamps. This time, I just passed in the URL and ran it through the @WayinVideo API. The best part about WayinVideo is that they are not just a product. They provide their services in a REST API. Which means you can plug this into your own workflows. For example: • auto-fetch your YouTube uploads • generate clips in bulk • push them to Shorts / Reels / TikTok • or plug it into an AI agent workflow (Claude Code, OpenClaw, etc.) The output isn’t just clips. It turns video into something you can actually work with. If you are building anything around video workflows, this is actually pretty interesting.

Have you noticed that new model launches don’t excite us anymore?

My laptop will be super hot.

@csaba_kissi Very briefly!

@Prathkum Did you already test it? Seems they try to increase the version number as soon as possible.

4.6 === 4.7

@MaMoMVPY Yes. LLMs made it obvious who is better.

It is not only coding - it is in everything. The specialist becomes a lot better with LLMs.

Unpopular opinion: AI tools like Claude Code actually widen the gap between junior and senior engineers. Top engineers can spot AI mistake in seconds because they know how the system should work. A junior who lacks that foundation can't understand what the AI generates.

@mvrckhckr Not just programming, if you use AI while learning any skill, you are basically outsourcing your brain.

@Prathkum Only if the junior engineer doesn’t use AI to learn and actually become better.

@csaba_kissi 100%

@Prathkum It seems this gap size will increase as more robust apps are created.

We understand the decision but saying “open source is dead” is just that you don’t want people to see your vulnerabilities anymore. 😭

clickup.com/brain/agents Build your own custom agents here.

I built a custom agent inside my @clickup workspace. Tag the agent in the comment or task, paste a blog link, and it gives me tweet ideas.

@csaba_kissi I use ClickUp almost daily to manage my work.

🚨 Just dropped: ClickUp just made AI agents feel a lot more real. Their new Super Agents can be created from a simple prompt, then actually work across your workspace the way people naturally work: → assigned tasks → mentioned in docs, chats, and tasks → messaged directly → triggered to run automatically That’s the part I think matters most: This isn’t just “chat with an AI in a builder.” You create the agent once, then it shows up where work is already happening. And it’s surprisingly approachable. Even if you’ve never built an agent before, the setup looks simple: Describe the role, give it context, and let it start helping. What makes it more than a wrapper is the memory system: → recent memory for what’s happening now → working memory for the task it’s handling → long-term memory across your team and workflows → learns and improves from your feedback It can also connect across your tools with permissions-aware search, so it works with the same access rules your team already has. Most AI tools still feel like assistants you have to babysit. This feels closer to adding a new teammate. Link in comments.

Those who ship great things today without writing or reading code were the top engineers even before AI. Don’t get fooled into thinking AI will make you rich. It only amplifies what you already know.

@SaidAitmbarek In a broader way, yes. But I guess claude code, codex, and cursor serve the same set of audience.

@Prathkum I don't think CLIs vs. IDEs address the same market.

I am seeing a lot of posts on my timeline about Claude Code and Codex. Is the Cursor hype fading away?

@shikhr_ a blinking indicator in any text field.

@Prathkum Who's cursor?

Is Opus 4.7 around the corner?

I passed a long-form YouTube video into an API and it gave me back: • 10+ short clips • auto-animated captions • timestamped summaries • searchable moments • a full transcript with speaker labels All automatically. Normally, repurposing this kind of content is painful: Scrubbing timelines, finding highlights, clipping, writing captions, pulling timestamps. This time, I just passed in the URL and ran it through the @WayinVideo API. The best part about WayinVideo is that they are not just a product. They provide their services in a REST API. Which means you can plug this into your own workflows. For example: • auto-fetch your YouTube uploads • generate clips in bulk • push them to Shorts / Reels / TikTok • or plug it into an AI agent workflow (Claude Code, OpenClaw, etc.) The output isn’t just clips. It turns video into something you can actually work with. If you are building anything around video workflows, this is actually pretty interesting.