Neural Browser

The Neural Browser is a real web browser driven by an AI agent, running inside your Eyeleven OS. You give the agent a task in plain language — "find dental clinics in Guadalajara and collect their contact details" — and watch it browse live: the page it sees, the steps it takes, and the data it extracts, all streamed to your desktop window.

Starting a session

1. Open Neural Browser from the desktop.
2. On the Start a session screen, choose one of two modes:
   • New Agent Task — launch an AI agent to browse autonomously.
   • Browse Manually — take manual control of the browser yourself.
3. If you chose the agent, type your task where it says "Tell the agent what to do…" and press Send.

The window shows Waking the browser engine… while the session boots, then goes LIVE. Sessions are powered by the EyeNigma-1.0 engine.

The live view

The center of the window is the live page view — actual frames from the browser, updating in real time. Status is always explicit:

• Agent driving — click to take over: the agent is in control; clicking anywhere on the page hands control to you.
• Manual control — you are driving: your clicks and keystrokes drive the page. Toggle Agent ON / Agent OFF to hand control back.
• Agent idle — waiting for a task: the session is up but nothing is queued.
• Browser engine offline: the session ended or the engine is unavailable; use Retry or start a new session.

Steps, tasks, and extracted data

A side panel tracks everything the agent does:

• Steps — the agent's action log, filterable by All / Actions / Data / Errors. Watch it navigate, click, type, and read.
• Tasks — the queue of tasks you have sent.
• Extracted — leads the agent captured while browsing: name, company, email. Each card has a Send to CRM button, and Send all to CRM pushes the whole batch into your workspace Leads in one click. Sent cards are marked Sent.

CRM capture is what makes the Neural Browser more than a novelty: prospecting research turns directly into records your OmniForces can follow up on.

Credit cost

Each agent task costs 5 Neural Credits:

• The charge applies when the task actually starts — a task that fails to start is never billed.
• If your balance cannot cover a task, the browser refuses to start it and reports the billing limit instead of running for free and failing later.
• Browsing manually does not consume credits per action; only agent tasks are metered.

Tips

1. Be specific about the output you want: "collect name, email and phone for each result" captures cleaner leads than "look for clinics."
2. One goal per task works better than a compound mission — queue several tasks instead.
3. If the agent gets stuck on a page, click to take over, get past the obstacle manually, then switch Agent ON again.
4. Review the Extracted panel before sending to CRM — you decide what enters your pipeline.