Notte
Build and deploy reliable browser agents at scale
874 followers
Build and deploy reliable browser agents at scale
874 followers
One platform to build and deploy enterprise-grade browser agents. Managed infrastructure (sessions, proxies, identities, vaults) via a single API. Hybrid architecture combines deterministic scripts with AI reasoning for production reliability.
This is the 2nd launch from Notte. View more
Demonstrate by Notte
Launched this week
Record any browser task once and get production-ready code instantly with Demonstrate Mode.
Edit further your code in our Automation Studio with live browsers, deploy automation code as a serverless function, and schedule it to run autonomously. Managed sessions, proxies, identities, and vaults handle everything behind the scenes.
The fastest path from prototype to production in one unified platform.
Free Options
Launch Team / Built With
Notte
Hey Product Hunt! 👋 We're Notte (YC S25), and today we're launching the automation tools we wished existed when we started building browser automations.
Why we built this
Most real workflows need both deterministic scripts (for the reliable parts) and agents (for handling variation). Today you're forced to choose between:
Black-box agents that look magical until they silently fail
Fragmented DIY stacks where you juggle infra, orchestration, and app state across tools
There's a psychological gap too: when tools give you visibility and control, you debug and improve. When they hide the mechanism, you just assume "agents don't work" and abandon them. We wanted one environment where you can go from "I have a task" to "it's running in production" without context switching, and where building agents feels more like craft than simply prompting.
What the Console gives you
🎬 Demonstrate Mode – Just do the task manually once. Notte records every action and generates production-ready automation code. No prompts, no syntax guessing. Show your workflow, get editable code.
💬 Agent Mode – Describe what you want in natural language. The agent executes it in a live browser session; click "Map to Script" and it becomes editable code you own.
📝 Code Editor – Write automation scripts with a live browser beside your editor, AI code assistance, and rich debugging. See exactly what's happening as your code runs.
🚀 Zero-Config Deployment – One click to deploy your script as an API endpoint, with scheduling and cron. No infra setup.
Built-in agent tooling
Managed identities (auth, 2FA, account setup)
Secure vaults and proxy rotation
Session management and managed infra
Live debugging and execution logs
Who this is for
Devs (or vibe-devs) who need browser automations that actually work in production, whether it's data extraction, form filling, testing, or integrating with apps that don't have APIs.
We're a small team obsessed with developer experience and making agent tooling feel empowering instead of opaque. Would love your feedback in the comments!
👉 Try it free: http://console.notte.cc
– nottelabs team:) 🌸
When workflows change later like a small UI update or an extra step added by the site how easy is it to update just that part of the recorded flow without redoing everything from scratch or risking downtime?
Notte
@why_tahir the cool thing is you don't redo anything. Wrap the script with AgentFallback, then if the selector breaks, the agent handles that step dynamically and the rest of the flow continues. Fix it later when you have time or let the agent keep covering it (you get the adaptability of agents without paying for them on every run). You can edit the code in our built in IDE when you get round to it.
Notte
biased but this has genuinely been super useful + intuitive for me 🔥 can pretty much automate anything I can think of
Notte
@samatnotte 🚀
Helploom
Great product!
Notte
@preetmishra thanks preet!
Notte
@preetmishra thanks!! 🏆
Product Hunt
Notte
@curiouskitty Use agents to discover, deterministic code to execute, and AgentFallback as your safety net.
The longer answer:
Black-box agents fail silently because you can't see what they're doing or why. Brittle scripts fail loudly because they're hardcoded to a specific UI state. Our approach:
Agent Mode — Use it to prototype and discover the flow. Let the agent click around, figure out the selectors, handle the edge cases etc. Great for exploration/prototyping.
Demonstrate Mode — You do the workflow once in a real browser, we record it and generate deterministic code. You get Playwright-style code that runs the same way every time. This is for people who know the workflow but don't want to write code.
AgentFallback — This is the key. Wrap any deterministic block with an agent fallback. If the selector breaks because the site changed, the agent picks it up dynamically instead of failing. You get the reliability of scripts with the adaptability of agents, but only when needed.
TLDR the philosophy is: agents should be scoped and supervised, not autonomous. Bootstrap fast, convert to deterministic, let agents handle the exceptions.
Notte
We believe web agents are a great way to bootstrap an automation but they should not be used blindly in production settings especially when the stakes are high. That's why we propose to convert a bootstrapped automation (or a demonstration) into deterministic code that you can audit + an agent as a fallback in case the UI changes to make sure the automation does not break. You get the best of both worlds
Notte
@giordano_lucas couldn't have said it better 🤝
A lot of browser automation tools fall apart right after the demo. You can record something or prompt an agent but turning that into code you trust in production is the hard part. Demonstrate Mode looks like it’s trying to close that gap by letting you show the task once and then work directly with the generated code. That focus on visibility and control is what usually decides whether an automation actually ships.
Notte
@elin_sjoberg well put - add AgentFallback to the deterministic script you Demonstrated, and you win on cost, reliability, and speed
I have used Notte for quickly building web agent automations. Excited to use Demonstrate Mode to iterate faster!
For quite some time, I believe web or UI agents that allow user to demonstrate use and AI auto correct are the ones that have highest reliability while minimizing maintenance cost. Notte's feature set is very close right now.
Notte
@raimo_tuisku thanks Raimo! and I agree, that balance of deterministic speed and cost with agent adaptability is where we're headed. Demonstrate Mode gets you there fast then add AgentFallback around you script to keep it running when things break.