Automation & the assistant

OpenLeads v3.5 can run your outreach for you: describe a campaign in one line, and let your own machine find, draft, and send it at the right time, with every message landing in your Sent folder like you typed it yourself.

#The assistant (natural language → a configured campaign)

Tell OpenLeads what you want in plain English:

bash

openleads assistant "send 50 emails to fintech founders for my SaaS at 9am"

It parses the audience ("fintech founders"), the count (50), what you're pitching ("my SaaS"), and the send time (9am), then finds the leads, drafts a personalized email for each, and previews everything as a dry run. Nothing is sent until you say so.

You can do the same conversationally inside openleads chat: just type the request. When you ask to send or schedule, it routes to the assistant; when you ask to find, it searches.

It works fully offline with a rule-based interpreter (no key required). If you set OPENROUTER_API_KEY (a free model is fine), messier phrasing is understood by the LLM, which fills the same validated action schema.

#What "reach" means for a campaign

A campaign drafts every safe (deliverable) lead plus high-confidence risky ones, those scoring ≥ 55% on the calibrated likelihood. On the common case where outbound port 25 is blocked, a healthy authenticated corporate domain with a common pattern scores ~62%, so you still get a real, sendable batch instead of an empty list. Catch-all guesses are deliberately kept below that line.

#On-device scheduling

Give a recipe a schedule and arm your machine once:

bash

openleads recipe add miami "agencies in Miami" --every "weekdays 9am" --send
openleads schedule on              # launchd (macOS), cron (Linux), Task Scheduler (Windows)
openleads schedule status          # is it installed, and what runs next?
openleads runs                     # what ran, how long it took, what failed

The agent wakes every 15 minutes and runs whatever is due: recipes, watchers and sequence follow-ups, always within your warmup/daily cap. Schedules, run history and failure alerts are covered in Timed runs. Run a cycle by hand anytime with:

bash

openleads drip            # dry-run preview
openleads drip --live     # actually send

#Send-time intelligence

Sends are paced like a human, not blasted: weekdays only, inside business-hour windows (08:00–11:00 and 13:00–16:00 by default), spaced apart with jitter, capped per day. Recipient time zones are inferred offline from country/city so your email lands in their morning, wherever they are.

#Emails show up in your Sent folder

Mail sent over raw SMTP normally never appears in your provider's Sent folder (Gmail and Workspace are the exception: they journal it for you). OpenLeads fixes this for everyone else by saving each real send to your mailbox's IMAP Sent folder.

Control it with the save_to_sent setting:

Value	Behavior
`auto` (default)	Save to Sent for every provider except Gmail/Workspace (which already do).
`always`	Always save (use if your Gmail setup doesn't journal sends).
`never`	Don't touch IMAP.

bash

openleads config set save_to_sent always

This needs IMAP configured (it falls back to your SMTP credentials). It's best-effort and fully isolated, so a Sent-folder hiccup can never fail an actual send.

#Safety rails (unchanged)

Dry-run by default everywhere, add --live to send for real.
Suppression-aware: never emails anyone who bounced, unsubscribed, or was already contacted.
Warmup-capped: a fresh mailbox ramps up gradually instead of blasting day one.
Plain text, a real List-Unsubscribe header, and a polite opt-out footer.

See Deliverability for the sender-side reputation checks, and Sending for mailbox setup.