Skip to content
AgentCorp
// Guides

User guide

How to get real work done once you are set up. AgentCorp is designed to feel like directing a small, capable team — you say what you want, the agents figure out the how, and you stay in control of anything that matters.

Talk to Alex, not to a menu

Alex is your single point of contact. You do not choose which specialist to invoke — you describe the outcome you want and Alex orchestrates the rest, pulling in Sam, Riley, or Taylor behind the scenes and coordinating their work. Write the way you would brief a competent assistant: the goal, any constraints, and how you want to see the result.

A well-formed request
“Pull the three deals in my CRM that haven’t moved in two weeks, draft a short nudge email for each, and show them to me before anything sends.”

That one message touches the CRM (Sam), composes outreach, and defers sending to your approval — all from a plain-language brief.

The specialists

You will see each specialist’s name attached to the work they do, so you always know who did what.

  • Alex — executive assistant and orchestrator. Plans, delegates, and reports back. Your default interface.
  • Sam — sales and CRM. Pipeline hygiene, lead research, outreach drafting, follow-ups.
  • Riley — finance. Invoices, expense summaries, reconciliation prep, cash-flow questions.
  • Taylor — marketing. Content drafts, campaign ideas, social copy, positioning.
  • More specialists are coming soon; you will see them appear on your team as they launch.

Approvals — your safety checkpoint

Any sensitive or outbound action stops for a human approval before it runs. When an agent wants to send an email, message a contact, or take another action that leaves your account, you get an approval card showing the real recipient, the subject, and the full message body — exactly what will be sent.

  • Approve to let the action proceed as shown.
  • Edit first if the wording or recipient needs a tweak, then approve.
  • Reject to cancel it; the agent will adjust or ask what you want instead.

Read-only work (reading your inbox, summarizing a document, searching your CRM) does not require approval. See AI best practices for how to decide when to keep a human in the loop.

Briefings and the situation room

Rather than pinging you constantly, agents batch what they have done and surface it as briefings — short digests of completed work, drafts waiting on you, and things worth your attention. The situation room is your at-a-glance view of everything in flight across the team: active tasks, pending approvals, and recent results, in one place.

Notifications and channels

You can work with your team from more than the web app. Connect WhatsApp to message Alex from your phone, or use the desktop app. Notifications — a task finished, an approval is waiting, credits are running low — reach you on the channels you have set up so nothing stalls because you were not at your desk.

Tips for good results

  • State the outcome, not the keystrokes. “Get me ready for my call with Acme” beats a list of micro-steps.
  • Give context once. Anything you upload to your knowledge base is available to every agent, so you do not repeat yourself.
  • Ask for a plan first on big tasks: “Outline your approach before you start.”
  • Verify anything consequential. Agents are capable but not infallible — check facts, figures, and names before they matter.