Is my AI agent shared with other users?

No. Every LeadClaw user gets a dedicated AI agent running on its own isolated VPS. Your agent's data and outreach never mix with any other user's account.

What is my agent doing when I'm not logged in?

Your agent runs 24/7 regardless of your login status. It checks your inbox every 2 minutes, sweeps for new leads every 4 hours, works the outreach queue every 60 seconds, and sends a heartbeat to your dashboard every 2 hours.

Which AI models does my agent use?

By default, Gemini Flash handles generation and reasoning while Gemini Flash-Lite handles quick classification tasks like sentiment analysis. Claude Haiku and Sonnet act as fallbacks, Sonnet for complex reasoning and Haiku for simple classification, only stepping in when Gemini is unavailable.

Does my agent work during weekends and holidays?

Yes, your agent runs 24/7 on every plan. Outreach avoids 10 PM to 7 AM in the lead's local timezone as a courtesy, but there's no weekday-only or business-hours restriction.

How Your Agent Works

Your AI agent runs nonstop on a server that's yours alone. It works through six loops that each handle a different job, and it picks the right AI model for each one automatically.

Dedicated Infrastructure

Your agent runs on its own virtual private server (VPS). It's never shared with another customer.

That gets you three things:

Privacy: your data never mixes with anyone else's
Performance: consistent resources, not a shared pool
Isolation: a problem on one agent can't spill over to another

The 6 Loops

Your agent runs six loops at once, each on its own schedule.

Loop 1: Inbox Monitor (every 2 minutes)

Checks for inbound email replies from leads. Classifies each response by sentiment and updates lead status automatically.

Loop 2: IPC Watcher (every 5 seconds)

Handles real-time communication with the LeadClaw web app: processing commands, reporting status, syncing data.

Loop 3: Task Scheduler (every 60 seconds)

Manages the outreach queue. Picks leads to contact, drafts emails, schedules sends, and handles follow-ups.

Loop 4: Awareness Sweep (every 4 hours)

Looks for new leads that match your active goals, and keeps an eye on what needs your attention.

Loop 5: Owner Message Monitor

Your messages get pushed to your agent instantly. A backup check every 5 minutes catches anything missed during a restart.

Loop 6: Heartbeat (every 2 hours)

Reports your agent's health to the dashboard: uptime, memory usage, active work, error rate. This is what makes the status pill on your dashboard pulse when your agent is active.

AI Models

Your agent uses more than one AI model, and it routes each task to whichever one fits best.

Gemini Flash: the default for most of the work. Drafting emails, running goals, researching leads, writing reports.
Gemini Flash-Lite: cheap and fast, used for classification. Sentiment analysis, lead scoring, that kind of thing.
Claude Sonnet: the fallback for complex reasoning when Gemini is unavailable. Simple classification falls back to Claude Haiku instead.

Gemini handles almost everything, day to day. Claude only steps in when it has to.