How AI Agents Run Outreach In AutoEmails | Playbook
The exact procedure an AI agent follows to build and launch cold outreach in AutoEmails: businesses vs decision makers, questions to ask, variables, writing r
Why agents use AutoEmails
Normal mailboxes can't send cold outreach at volume. AutoEmails gives your agent finding leads, finding emails, warmed sending, daily allowances, reply tracking and unsubscribe handling — the agent just decides and writes.
1. Businesses or decision makers?
Businesses: companies found by scanning Google Search and Maps for a sector and place. Emails come from the company website via the built-in email finder. Free. {{first_name}} is usually empty, so 'Hi {{first_name}},' becomes 'Hi there,'.
Decision makers: named people by job title and seniority, with verified emails. Uses Decision Maker credits. Needs the Decision Makers integration or managed credits.
Rule: sectors and places only → businesses. Roles named (MD, Operations Director, Head of Procurement) → decision makers. Call audience_options to see credits.
2. Ask before building
Call outreach_setup_questions and ask the user, in one message, anything you can't infer: sectors, towns, how many leads per scan, how often to scan, days and hours, sender, number of emails and gaps, who writes the copy, test recipient.
3. Build the draft
Call create_outreach_campaign with the answers. Pass steps with your own subject and body to send them word-for-word, or omit steps to let Leo write. The campaign, its emails and the lead scan are created exactly as the web wizard does, so they show up normally in AutoEmails (scan switched off until go-live).
4. Writing rules
UK English and £. Subjects in Title Case; {{first_name}} only at the start (it's moved there automatically).
Outreach is plain text with the sender's signature (attached automatically) — don't write a sign-off name in the body.
Use variables: {{company_name}}, {{first_name|there}}, {{sector}}, {{booking_link_1}} (always works). Create your own with set_variable, e.g. {{offer_line}}. Call list_variables for the full list.
5. Test first — always
Call preview_email, then send_campaign_test (defaults to the workspace sender). Show the user and wait for approval. Live sends are refused unless this exact copy was tested in the last 24 hours.
6. Go live
Call activate_outreach_campaign. It returns a summary; show it and ask 'Confirm?'. After an explicit yes, call again with the confirmation_token. Pause any time with pause_campaign.
7. Report
get_campaign, list_replies (Inbound), list_meetings and client_report give results in plain English for clients.
Integrations
list_integrations shows what's connected. connect_integration saves an API key (Decision Makers/Apollo, Calendly, OptiMonk, WooCommerce) after the user confirms; keys are never shown back. Shopify and Google need a browser sign-in — the agent gives you the link.
About AutoEmails and Leo
AutoEmails is an AI email marketing platform built around Leo, an autonomous AI email marketing agent. Leo analyses your business, finds the right audience, writes and designs emails, launches campaigns and reports on replies, meetings, orders and revenue.
The platform covers three engines in one account: Outreach, which finds new companies or decision makers and emails them with personalised plain-text messages and follow-ups; Marketing, which sends designed HTML newsletters and campaigns to your own contacts; and Automations, which run triggered email journeys such as welcome series, abandoned cart recovery and re-engagement.
Leo works autonomously within guardrails. It can suggest and launch campaigns, resend to non-openers, expand targeting when a lead scan falls short and pause sending when deliverability signals such as bounce or spam-complaint rates rise. Every action is logged with an evidence trail so you can see what Leo did and why.
AutoEmails connects to Shopify, WooCommerce, WordPress, Calendly, Google Analytics, Google Search Console and custom e-commerce stores, so campaigns can be measured on real outcomes: replies and booked meetings for lead generation, and orders and revenue for e-commerce.
Pricing is capacity-based in British pounds, starting with a free plan. You can start free at autoemails.com without a credit card, let Leo scan your business and review the first campaign before anything is sent.
More from Leo actions
- AutoEmails MCP Reference | Every Agent Action — Every action AI agents (ChatGPT, Codex, Claude, Cursor) can take in AutoEmails via MCP, with inputs and which ones ask you to confirm.
- AutoEmails MCP For Cursor | Setup Guide — Use AutoEmails from Cursor's agent through MCP: build campaigns, read results and send tests from your editor.
- AutoEmails MCP For ChatGPT | Custom Plugin Setup — Step-by-step: add AutoEmails as a custom MCP plugin in ChatGPT so your GPT agent can build outreach, send tests and run campaigns through AutoEmails' bulk sen
- AutoEmails MCP For Codex | Plugin & config.toml Setup — Set up AutoEmails as an MCP server in OpenAI Codex — desktop plugin with OAuth or config.toml with a workspace key — and run outreach campaigns from your agen
- Email Marketing For AI Agents | AutoEmails — Let Claude, ChatGPT, Gemini, Copilot or Cursor run your newsletters, HTML designs, CRM sync and unsubscribes — with you approving every send.
- Run Email Marketing From Your AI Agent | AutoEmails — AutoEmails is the safe email layer for AI agents. Ask ChatGPT, Claude, Cursor or Gemini to build newsletters, push HTML designs, send tests and sync CRM contacts — live sends wait for your approval.
- How AutoEmails Works — AI Email Marketing Agent Explained — See how the AutoEmails AI agent finds leads, writes personalised emails, launches campaigns and runs automations for you — outreach, marketing and automation in one platform.
- Pricing Plans from Free to Enterprise | AutoEmails — Transparent pricing for every stage of growth. Start free, scale as you grow. No hidden fees, no audience-based billing. Cancel anytime.