Browser-use Hackathon ⚡ Delhi Edition
Delhi, we are bringing the Browser-use Hackathon to you. ⚡
Spend a Saturday building AI agents that can research, test, monitor, book, shop, and complete useful work across real websites.
Come with an idea, join a team, or build solo. You do not need prior browser-automation experience. We will begin with a practical Browser Agents 101 walkthrough so everyone can start building quickly.
🎯 The theme: Delhi Edition
Build an agent that solves a real browser workflow. Pick a lane:
🧑🚀 Personal operator: automate repetitive tasks such as scheduling, forms, account management, and recurring web workflows
🔎 Research and intelligence: collect information across websites, compare results, monitor changes, and return structured findings instead of a pile of tabs
🧪 Developer and testing: test a product like a real user, reproduce bugs, inspect live interfaces, and turn stable checks into repeatable workflows
🛒 Commerce and bookings: search, compare, prepare carts or bookings, and create safe approval points before payments or other sensitive actions
📡 Monitoring and operations: watch dashboards, listings, prices, availability, or business-critical pages and act when something changes
🃏 Wildcard: if an agent operating a real browser makes the workflow meaningfully better, it belongs here
⭐ Build on webcmd
webcmd is self-learning browser infrastructure for AI agents.
Your agent can explore an unfamiliar website, preserve what it learns, and turn stable workflows into deterministic CLI commands with structured output. Instead of reasoning through the same interface on every run, it can reuse a reliable command and return to the browser when something changes.
Explore once. Learn the workflow. Reuse the command.
Get started:
The webcmd team will be in the room to help with setup, browser workflows, adapters, and debugging.
🛠️ Bring your preferred stack
Use Claude Code, Codex, OpenClaw, Playwright, Browser Use, browser MCPs, local models, hosted models, or any APIs and tools that help your agent complete the job.
⚖️ How we judge (100 points)
🟢 Live reliability, does the workflow complete on stage: 30
💡 Usefulness, does it solve a browser task someone genuinely wants automated: 25
🧠 Technical depth, how well does the agent reason, recover, and reuse what it learns: 20
✨ Creativity: 15
🎤 Demo and storytelling: 10
🚨 Two hard rules
Your demo must run live or use a screen recording captured from a real execution. We reward agents that actually complete the workflow.
Build responsibly. Use your own accounts, respect platform terms, and keep a human approval step for payments, messages, submissions, deletions, or other sensitive actions unless you are working in a safe sandbox.
⏱️ Format
🟠 Saturday, August 8, 2026 · 10:00 AM to 5:00 PM
🟠 We open with a 30-minute Browser Agents 101 walkthrough, including a quick webcmd demonstration
👥 Come solo or form a team of up to four
🎒 What to bring
A laptop, your preferred model or API keys if you already have them, and one browser workflow you would love to stop doing manually.
🙌 Who should come
AI-agent builders, open-source contributors, automation engineers, developers, students, founders, local-model enthusiasts, and anyone curious about giving agents useful access to the real web.
No prior webcmd or browser-automation experience is required. The opening walkthrough has you covered.
📍 Delhi · Hosted by webcmd
Spots are limited to keep the room practical and high-signal. Hit RSVP and bring a friend who builds. 👇