What Is an AI SDR?

An AI SDR is software that does the first-touch work of a sales development rep: it finds prospects, writes outreach, sends it and handles the follow-up with little human input.

Key takeaways

• An AI SDR automates prospecting and first contact. A human still owns the offer, the targeting and the live conversation.
• Its output is only as good as its input. Wrong titles, stale records and bad emails produce wrong messages at machine speed.
• An AI SDR is not a data source. It needs a separate feed of people, work emails and company context.
• On LeadOcean the feed is one API or MCP server. Sizing a list is free and each person returned costs 1 record.

What it is

An AI SDR is an agent, usually built on a large language model, that researches accounts, picks contacts, drafts personalised outreach and sends and follows up on it.

The human version of the role books meetings for account executives. A rep builds a list, finds the right person, writes the email, sends a sequence and hands over replies that show interest. An AI SDR does the same steps in software.

Most products split the job into three parts: a data source, a model that writes, and a sender. Vendors bundle them to different degrees. Some ship all three. Others ship only the writing layer and expect you to bring the contacts and the mailbox.

That split is the useful way to judge one. Ask where its contacts come from, who sends the mail, and who reads the replies.

How it works

An AI SDR runs a loop. It takes your ideal customer profile (ICP), turns it into a list, and works the list one person at a time.

1. Take the ICP. You give it titles, company size, region and industry. This is the one input only you can supply.
2. Build the list. The agent searches a contact database for people who match the ICP.
3. Check the contact. It reads the email status and drops addresses that would bounce.
4. Research the account. It reads the company's site, news or job posts to find a reason to write.
5. Draft the message. The model writes a short email that uses that reason.
6. Send and follow up. A connected mailbox sends it, then sends the next steps on a timer.
7. Route replies. Positive replies go to a human. The agent stops writing to that person.

Worked example, with placeholders. Your ICP is VPs of Sales at US companies with 201-500 employees. The agent pulls Jane Doe at acme.com, sees her email_status is verified, reads that Acme posted three sales roles, and drafts two lines about hiring ramp time. It sends the email on Tuesday and a follow-up four days later. Jane replies, and the agent hands the thread to an account executive.

AI SDR vs sequencer vs human SDR

People confuse an AI SDR with an email sequencer, because both send timed emails. The sequencer sends what you wrote to who you picked. The AI SDR also picks and writes.

Pick a sequencer when your list and copy are good and you only need delivery. Pick an AI SDR when research and writing per contact is the bottleneck. Keep a human for calls and for anything with a real budget at stake.

For a ranked list of tools, see the best SDR tools for prospecting.

When it matters

You have a small team and a large market

One founder or one rep cannot research 2,000 accounts. An AI SDR covers the long tail that would otherwise get no outreach. Keep the list tight, because a large wrong list is worse than no list.

Your message depends on a fact about the account

A hiring signal, a new office or a technology on the site makes a good first line. An agent can read those facts at scale, if it has them. Give it company data as well as people data.

You already have a sender and want better input

Many teams have a mailbox setup that works. They lack a clean list. Here the AI SDR matters less than the data feed behind it, which is why the data an AI SDR needs is the deciding factor.

You are about to hand your sending domain to an agent

An agent that sends fast and fails fast can burn a domain in a week. Gate sends on email status, start small, and watch bounces. LeadOcean does not refund or credit bounced emails, so read email_status first.

How LeadOcean handles it

LeadOcean is the data feed, not the SDR. It has no sequencer, inbox or dialer. Your agent calls the API or the hosted MCP server at https://api.leadocean.io/mcp, signs in with OAuth, and gets people, work emails, phones and company records.

Each step of the loop above maps to a call:

• Size the ICP: POST /v1/people/search?count=true with limit=1, or leadocean_count_leads. Free.
• Pull people: POST /v1/people/search. 1 record per person returned.
• Check the contact: read email_status on each row. The emailStatus filter takes verified, catch_all_valid and catch_all by default.
• Add company context: POST /v1/companies/enrich, 1 record per company.
• Add a phone for call tasks: GET /v2/people/phone, 1 record per call.

We ran leadocean_count_leads on 1 October 2026 with jobLevel VP, jobFunction Sales & Business Development and country US. It returned 115,336 mailable people (the default filter: verified, catch_all_valid and catch_all). The call cost 0 records. Over REST, meta.total is capped at 100,000, so the call below returns 100,000 for this filter set.

curl -X POST "https://api.leadocean.io/v1/people/search?count=true&limit=1" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"jobLevel": ["VP"], "jobFunction": ["Sales & Business Development"], "country": ["US"]}'

Drop count=true and raise limit (up to 100) to pull the people. Page with cursor. The Exports page in the app at app.leadocean.io builds the same filter and shows the record price before you start.

Pricing is two plans: Free (1,000 records, one-off, no card) and Pro at $499 a month. See pricing and the AI SDR data use case.

FAQ