No fine print, no dodging. If we haven't covered it, the contact form is right below.
Are these real AI agents, or canned responses?
Real agents. Each one runs a live model with tools — it crawls your site, reads your analytics, drafts copy, and reasons about what to do next. The timeline on this page is a faithful render of what the console actually shows.
Which actions still need my approval?
Money — and only money. A charge, a refund, a payout, connecting a live payment key, or real ad spend pauses for your yes, with the exact amount shown. Everything else — drafts, a published article, a non-paid outreach email, a deploy — the fleet ships on its own. The money gate is enforced in code, and automatic safeguards (a kill switch, suppression/opt-out honoring, and anti-injection on anything the agents read) run on every action without a prompt.
What can the agents actually do?
SEO audits, content drafts, social calendars, email campaigns, ad planning, analytics digests, brand-voice checks, and outbound prospecting. They research, write, and plan. They don't pretend to be human, and they don't act on the outside world without your yes.
How is this different from ChatGPT or a single AI assistant?
One assistant gives you a blank box and waits. ipop gives you a standing department — eight specialists in their own channels, working in parallel, with a shared memory and a human approval layer. It's a team, not a tab.
Is my data safe? Can one customer see another's work?
Every request is scoped to your workspace; tenants are fully isolated. Each agent only receives the credentials its job needs, and outbound network access can be locked to an allowlist. The honest details — including what we haven't built yet — live on our security page.
What does it cost, and can I try it first?
Start free, no card. Paid plans run from $49 to $499 a month, and agent compute is billed against a cap you set — we never cross it. You can change or cancel any time from the billing screen.
What happens if an agent gets something wrong?
It surfaces as a draft, not a live change, so a mistake costs you a glance, not a cleanup. You send it back with a note, the agent revises, and the whole exchange is logged so it learns the preference for next time.
Do I need to be technical to use it?
No. If you can send a Slack message, you can brief an agent — you @mention them by name and say what you need. There's no setup call, no Gantt chart, and no prompt-engineering homework.
Can I keep my own tools and just add the agents?
Yes. The agents work alongside what you already use — they draft into your channels and hand off the finished thing. You stay in control of where it goes.