Hey PH
We're launching Rover on Feb 25th, but the preview is live right now and we want your honest takes before we go big.
What it is: One script tag on your website your users get an AI agent that takes real actions inside your UI. Clicks buttons, fills forms, runs checkout, guides onboarding. Through conversation.
<script src="https://rover.rtrvr.ai/embed.js"></script>
Cloud Run powers the backend behind Retriever’s Chrome extension, cloud platform, API, MCP and WhatsApp. We deploy our agent services on Cloud Run. Here’s how the pieces fit together:
On-device execution: DOM parsing and browser actions run in the Chrome extension. Our Cloud Run backend handles model requests and coordinates the agent loop.
Cloud execution: Cloud Run services dispatch tasks to isolated Chrome sessions, track progress and return results. Users can run tasks in parallel.
Scheduling: Cloud Tasks dispatches scheduled runs, with Firestore tracking their state. Execution IDs, worker leases and concurrency policies help prevent duplicate or overlapping runs. Retries have explicit limits and backoff.
WhatsApp and webhooks: Queued workers handle the task and send updates back to the chat.
Scaling: We keep latency-sensitive routes warm, allow event-driven workers to scale to zero, and set concurrency, memory and timeout limits for each workload.
congrats @bhavani_kalisetty @arjun_chintapalli - it can send outreach on its own, great! until it sends 200 dms you didn't want. is there an approval step or a daily cap before anything goes under my name?
@bhavani_kalisetty @vlad_popov3 The agent already knows about LinkedIn limits and will ask you back for confirmation if needed!
Hey Product Hunt, Arjun here, the other cofounder!
A fun fact about this launch: Retriever helped run it. I had Claude hand our outreach to Retriever over MCP, and it sent our DMs on X and filled out our directory submissions while we worked on everything else. That’s exactly the work we built it for: tedious, multi-step, and spread across sites that have no API.
A few things you can do with it today:
Free Mode: run tasks for free, funded by small sponsored ads
Cloud schedules: have a task run every morning, even with your laptop closed. Our entire cloud browser infra is hosted on Cloud Run, shout out to @Google Cloud for amazing resilient infra!
Cookie sync: log in once in Chrome, and cloud runs stay signed in
Site skills: it remembers how a site works, so repeat tasks get faster and cheaper
MCP: hand web tasks to Retriever from Claude, Codex or Cursor
I’m here all day. Drop a task you’d want automated in the comments, and I’ll run it and reply with what happened, good or bad. Honest feedback on where it breaks helps us most.
Arjun
@arjun_chintapalli impressive to see so many features in one AI agent which use case would you be recommend trying first ?
@dipanshu_kushwaha5 The killer use case everyone wants is to automate LinkedIn outreach in your browser completely for free, just prompt agent to:
- send connection requests
- sync DM's to CRM
- apply to jobs
from 35k+ users already, what are the most common tasks that you noticed people delegating in the "let's see what it can do" phase?