Inbound is flat and there is no salesperson. Run outbound from the window you already have open.

The short answer for a SaaS founder

LinkedIn outreach for SaaS founders works as a small daily habit, not as a project. If you already spend the day in Claude Code or ChatGPT, you can add the GTM Bud connector, build one campaign in one session, and then keep it alive in about 30 minutes a day: answer replies, read the numbers, make one change.

This page is for a technical founder of a B2B SaaS with fewer than 15 people and nobody in a sales role. Inbound has gone flat, and outbound has to fit between releases. If you want the step by step build of a first campaign, read a technical founder’s first LinkedIn outbound. This page answers a different question: is the daily load small enough to carry.

What the 30 minutes hold

The slot has three parts, and the order matters.

Minutes	Part	What you do
0 to 15	Replies	Read new replies, edit the answers your AI wrote, approve the good ones
15 to 20	Numbers	Ask for the campaign results and read them with the as-of time
20 to 30	One change	Fix one thing: a criterion, an offer, a playbook rule. Never three

Replies come first because a buyer who answered on Tuesday should not wait until Friday. Numbers come second because they take two minutes. The single change comes last because it is the part you can skip on a bad day.

The 30 minutes are a planning estimate for one campaign on one LinkedIn account. A campaign that gets many replies costs more review time, and that is a good problem.

Three moments this fits

The flat quarter. Signups stopped moving in July. It is now October and the plan says “more content”. Outbound lets you start conversations this month, with buyers you choose.

The week after a release. You shipped on Thursday. Friday morning is the two hour build session: audience, preview, offer, messages, launch.

The investor update. The update is due on the last day of the month, and you want one honest paragraph about outbound. A per-campaign result with a date on it is that paragraph.

A Monday in the slot

This is a session in Claude Code. The tool names are the real ones. The people and the company are invented, and no result numbers appear because they would be yours.

you    New replies on the audit-log campaign?
agent  list_replies
       3 replies. 2 need an answer.

you    Open the first one.
agent  get_conversation
       Lead: Head of Security at a 90-person fintech.
       "We build our own audit trail. What breaks at our size?"
       Draft answer, written from the campaign playbook:
       "At about 100 people the usual break is export: an auditor
        asks for 12 months of events and the query takes a day.
        Is that request on your calendar this year?"
agent  review_reply
       Shown for your review. Nothing is sent.

you    Change "the usual break" to "what broke for our first
       customers". Then send it.
agent  send_reply
       Sent the approved text.

you    Label the second one, it is a no.
agent  classify_reply  (not_interested)

you    Numbers.
agent  get_campaign_results
       Results returned with the as-of time.

you    Stop new first messages until Thursday. Release week.
agent  set_sending_mode  (mode: followups)
       Applied at once. Follow-ups to earlier messages continue.

A worked example: the build session

The company here sells an audit-log API to B2B software teams. It has nine people. The founder, Ines, builds the campaign on a Friday morning.

The audience sentence. “Heads of security and engineering managers at B2B fintech companies with 50 to 300 employees in the United States and Canada.”

What came back. Her AI wrote five criteria from the sentence. GTM Bud passed four and flagged one, “uses a homegrown audit trail”, because a profile or a company page cannot show it. The suggested rewrite was “engineering team that publishes about compliance or SOC 2”. Ines made it a nice-to-have, not a must-have.

Two previewed people.

• Rahul M., Head of Security at a 120-person payments company in Toronto. Marked Good fit.
• Kate L., Engineering Manager at a 2,000-person bank in New York. Marked Not a fit: a bank, far above the size range, with a procurement cycle a nine-person company cannot survive.

The seller profile of the draft. The company website and the founder’s personal LinkedIn profile. GTM Bud reads both.

One value offer. “A written review of your current audit events against the SOC 2 evidence list, with the gaps marked.” A value offer is something free and useful that the campaign gives each lead. The AI wrote it from the GTM Bud instructions and the product documentation, and GTM Bud checked and saved it.

The launch. The readiness check listed one open item: no sending account chosen for the campaign. Ines fixed that on gtmbud.com, read the launch preview with the first message of each channel, read the full sequence, and confirmed.

Five prompts for the daily slot

• “Start my outbound slot: list the replies that need an answer, oldest first, and stop before you write anything.”
• “For this reply, write an answer a founder would write: one technical detail from our docs, one question, no sales words. Show it to me.”
• “Give me the campaign results with the as-of time, and tell me only what changed since the note I saved yesterday in outbound/log.md.”
• “Three leads asked if we support SSO. Propose one new rule for the campaign reply playbook that covers it, and show me the exact text before you save.”
• “It is release week. Tell me what follow-ups only would change on this campaign, then apply it.”

Three objections founders raise

“I will forget it by the second sprint.” Likely, if it lives in another product. Here it lives in the tool you already open every morning, and the opening line of the slot is one sentence. When you do miss a week, the campaign does not wait for you: it keeps sending until you change the sending mode. So put it in follow-ups only before a crunch, not after.

“I want it to run with no input from me.” This connector does not do that. Launch needs your confirmation of a preview. Your AI shows each reply for review, and GTM Bud instructs it to send only the text you approved. There is no scheduled run. For a founder that is the right trade: the first fifty conversations about your product are research, and an unread answer to a buyer is a poor way to do research.

“LinkedIn is my personal account.” It is, and the messages go out from it under your name. Read the full sequence before you confirm the launch, not only the launch preview, which shows the first message of each channel. A LinkedIn reply that went out with a mistake can be edited or deleted in the GTM Bud Inbox for 60 minutes after it is sent.

What the connector does not do for a founder

• It does not book the meeting. It brings each reply into your chat; the calendar step is yours.
• It does not write to your CRM. If your AI has a CRM connector that you added yourself, it can copy a lead there. That is your AI’s work with your other connector.
• It does not take a CSV of your waitlist or your churned users.
• Three steps happen on gtmbud.com: connecting your LinkedIn account, choosing the sending account for the campaign, and setting the sending schedule. The 14-day free trial starts when you launch your first campaign.

Setup

In Claude Code, the setup is one command, shown on GTM Bud for coding agents. In ChatGPT, follow GTM Bud for ChatGPT. If you have a GTM engineer or plan to hire one, the page on AI LinkedIn outreach for GTM engineers covers the multi-segment version of this work, and founder-led sales covers what to do with the conversations once they start.