ankit raj

The modern work stack is a distraction machine. Slack buzzes. A notification slides in. Your flow is gone — and you can’t even remember what you were doing 30 seconds ago. I wanted to fix this for myself. So my teammate built Focusd. Open-sourcing it today.

Director is our open-source framework for building video agents. Think of it as ChatGPT but for video. "Upload this and send me the highlights" "Find every mention of the product launch" "Create a 60-second recap" It just works. 1.3K+ stars. github.com/video-db/Direc…

We built a way to get off feeds. Instead of scrolling apps, a personal agent researches the internet for you and ships video briefings on a schedule. We call it Agentic Streams. No feeds. No ranking algorithms. Just signal.

We just released a new benchmark looking inside the "black box" of Gemini 2.5 reasoning for video understanding. Does "thinking more" always lead to better results? The answer is more nuanced than you’d think 💭

The internet is broken. Feeds decide what you see. Algorithms optimize for addiction. Truth gets buried under noise. So we built something else. "Agentic Streams."

“Bot joined the meeting” might be the most awkward sentence in software. It interrupts the room, waits to be admitted, then usually misses the screen anyway. I think the best meeting bot is no bot.

Your agents never see your meetings. So they can't help you live. They can't act after. All that context just disappears. call.md fixes that. 🧵

The Loom Tax: $12/seat/month for a folder of unsearchable MP4s. You record a bug walkthrough. Drop it in Slack. Two weeks later someone needs "the part where the API 500s." Nobody scrubs that video. The knowledge dies.

1/ Screen recorders are stuck in 2015. Subscriptions. Locked features. No access to your agentic system Your own recordings behind a paywall. We fixed that

We've got some incredible builders joining us for Perception 101 for your agents session 👀🎤 📍 Bangalore 🗓 Sunday, March 22 🕒 11:30 AM to 2:30 PM If you’re around, drop by. Would love to see you there. Secure your spot, link in the comments.

Running 10+ OpenClaw agents without monitoring is pure optimism. What happens when one goes rogue? We built CCTV for agents.

Our OpenClaw agent Arc had checkmate on the board and said nah, take my queen instead 😭♟️ Good news: we have CCTV for agents, so every game comes with a replay video.

Use it for: • agent monitoring • QA • security review • browser automation debugging • computer use dataset prep • shareable demos. github.com/video-db/openc…

@openclaw can use a computer. Now it can have CCTV. We built a monitor that records agent runs, streams them live, creates a sharable replay link, and lets you trigger alerts on weird behavior.

Hosting a few fun @videodb_io meetups this month in Pune, Bangalore, Tokyo and Delhi, and would love to have you all join. We’ll jam on agents with real-world perception, realtime AI workflows, and OpenClaw agents running YouTube channels. Keeping it light and practical. No slides, no theory dump. Just live demos, real discussion, and code you can take home. I’ll be at the Bangalore event on the 22nd and the Delhi event as well. Come by. Should be fun

Pair Programmer upgrade: 450 clones in last 14 days. people star but agents actually clone now an agentic skill ✨ not just Claude Code anymore works with Cursor, Codex, and more.. install with npx cleaner UI early signal: 95 stars 806 uniques in 14 days screen aware coding agents are coming

The blind spot nobody talks about. AI agents reason, write, browse. Continuous media? Still invisible. @videodb_io Skills: full video pipeline for AI coding assistants. Ingest, index, search, edit, automate.

I can now make memes on my phone 😄 checkout videodb skills.

@ashu_trv Trigger warnings sorted using a natural language command 😎

The only perception skill your agent needs 👇 npx skills add video-db/skills