AI agents for sales outreach
Research and first emails, approved before they send.
Agents that research prospects, draft the first email and classify the replies, with every send waiting for a person. Each run shows who was found, why, and what was written.
The job
Writing to prospects is repetitive. Choosing them is not.
Outbound looks like a perfect job for agents: research a company, write a first email, follow up, sort the replies. The writing and the bookkeeping repeat. Deciding who is worth writing to, and what goes out under your name, still needs a person.
How it runs
A pipeline of agents, with one gate.
One agent researches, another drafts, a third sorts the replies. Agents hand work to each other inside a pipeline, and the step that sends email waits for an approval, so nothing reaches a prospect that a person has not read.
- Research with sources. Web search and page reading, with the pages it used kept on the run.
- Drafts, not sends. The send step requires approval; approvers see the email as the prospect will.
- Replies sorted. Interested, not interested, needs a follow-up, out of office: classified, so a person answers the ones that matter.
- Cost per run. Token spend per agent, so a pipeline that grows does not grow its bill unnoticed.
What we learned
We ran our own outreach this way.
Agents are good at the writing and the bookkeeping; deciding who is worth writing to needed a person.
Step by step
A pipeline you can read end to end.
Research
An agent finds the facts on each company, with its sources.Draft
Another writes the first email from them.Approve
The send waits; a person reads each email as the prospect will.Sort replies
Replies are classified so a person answers the ones that matter.
Questions
What teams ask first.
Can it send email without anyone reading it?
Who picks the prospects?
Start from a template
Agents for this job, ready to install and adapt.
- EmailProspect Research AgentSearches the web for potential customers in a target segment, extracts contact info, deduplicates against the outreach sheet, and writes new prospects with status 'new'.Advanced setupGoogle Sheets
- EmailEmail Client AcquisitionSends personalised cold-outreach emails to prospects from a Google Sheet using per-segment copy. Marks each row as 'sent' after dispatch.Advanced setupGmailGoogle Sheets
- EmailLead Reply AgentSends personalised follow-up replies to outreach prospects classified as interested or needs_followup, then marks them in the sheet.Advanced setupGmailGoogle Sheets
- EmailOutreach Reply ClassifierReads the inbox, classifies replies to your cold outreach by intent (interested / needs follow-up / etc), and updates the outreach pipeline sheet.Advanced setupGmailGoogle Sheets
- EmailRe-engagement AgentFinds inactive users via your product's API and sends each a personalised re-engagement email. Input: { days_inactive, plan?, limit? }.Advanced setupGmail
What it runs on
The parts of AgentOS this job leans on.
Go deeper
How it works, in detail
From the blog
Case study · 9 August 2026What we learned from letting agents do our cold outreach
We built a pipeline of agents to find prospects and write to them. The writing worked. The finding did not. What we learned, and what we do now.
Guide · 16 August 2026Human-in-the-loop without the bottleneck
The difference between approving an action and answering a question, and how to bring a person into an agent's run only when it matters.
Pilot
Try it on one real workflow.
We build the agent with you, with an approval gate on the risky step, and report what it did.
Integrations
18 apps, your models, your APIs.
Connect GitHub, Gmail, Slack and more once, describe any HTTP API as a tool, and choose the model per agent.
Give one job to an agent this week.
Start from a template, gate the sensitive step behind your approval, and read the first run end to end.