# Founder-Led Sales: Stages, First Hire and LinkedIn Outbound

What founder-led sales is, its stages and when to hire your first sales rep – plus how a founder hands LinkedIn outbound legwork to an AI agent.

Founder-led sales is the stage when a startup's founder does the selling personally – finding the first customers, running discovery and closing deals – until the motion is repeatable enough to hand to a first sales hire. It is not the same as founder-led growth or founder-led marketing, which are about the founder's own audience as a growth channel rather than who does the selling.

> **The short version.** Founder-led sales is a stage, not a company type: the founder works out who buys and why, then makes it repeatable. Hand it over when the motion is documented and repeatable – practitioners put the bar at dozens of qualified prospects and a steady win rate, as the sections below show. On LinkedIn, an AI agent can take care of the legwork – list building, profile research, draft notes, and checking acceptances and replies – on your own account, waiting for your approval before anything is sent. The conversations and the closing stay yours, and what the agent's runs leave behind becomes the handoff pack for your first sales hire.

## What is founder-led sales?

The definition most pages build on comes from Pete Kazanjy's [Founding Sales](https://www.foundingsales.com/): "Founder Led Sales is the process by which startup founders discover, refine, and scale their product's initial sales motion. This includes the messaging, ideal customer profile, prospecting, sales materials, and more needed to close the first few dozen customers themselves, and then successfully hiring and onboarding early sales reps, before finally handing off this running sales machine to professional sales leadership."

Three things follow from it:

- **It is a stage with a handoff.** The founder sells first, then hires, then hands the running machine to sales leadership. The founder does not disappear from sales, though: Seth DeHart writes in [Growth Unhinged](https://www.growthunhinged.com/p/founder-led-sales-framework) that "at least one founder is the steward of revenue for the company. Forever."
- **It is a learning job before it is a volume job.** The first customers teach you the messaging and the ideal customer profile; the volume comes later, from people you hire.
- **The term is most associated with Kazanjy's book**, *Founding Sales*. Treat claims about who "coined" it with care – the pages that rank for it associate it with the book but do not name an inventor.

## The stages of founder-led sales

The most detailed stage model comes from a founder-led sales workshop [GTMnow](https://gtmnow.com/the-founder-led-sales-playbook/) ran with Pete Kazanjy and wrote up in March 2026:

| Kazanjy's stage | What is true | What to work on |
| --- | --- | --- |
| **Stage 0 – Zero Customers, Zero Motion** | No customers and no process yet | "validated ICP and problem-pain calibration, not revenue" |
| **Stage 1 – First Customers, No Playbook** | A few customers, nothing written down | "run 30 to 50 qualified prospects through your process, document every step, track your conversion rates at each stage" |
| **Stage 2 – Proven Motion, First Hire** | "A win rate of 15 to 25% on qualified pipeline, a defined sales cycle, and a clear map of the buying group" | Hire the first rep and check that the motion transfers |
| **Stage 3 – Repeatable Motion, Building Team** | "Two to four reps are running the motion consistently" | Bring in sales leadership; the founder's role "shifts from doing to enabling" |

Kazanjy also splits the whole journey into two modes – figuring it out, then scaling it – and the write-up names the classic error: "The most common mistake founders make in sales is entering Scale mode before they have actually finished Figure It Out mode." The first list he recommends is "a spreadsheet of 75 to 100 target accounts that match your ICP hypothesis", and a minimum viable sales motion is "somewhere in the range of 30 to 50 qualified prospects run through your process, with 10 to 20 converting".

Seth DeHart's framework in Growth Unhinged covers the same arc in six phases: finding early product-market fit, acquiring customers, the first sales hire(s), making it predictable, multiple sales hires and hiring a sales leader. All of these numbers are practitioner rules of thumb, not laws.

## How founders find a repeatable motion

Before anything can be delegated, the founder has to find out what works. That is a loop, run with real buyers, and each turn should leave something written down:

| Step | What the founder does | What it leaves behind |
| --- | --- | --- |
| 1. Hypothesis | Picks the customer and the problem to test | A one-paragraph ICP and the problem statement |
| 2. First conversations | Starts with warm introductions, then reaches out to people who match the ICP | A list of conversations and where each one came from |
| 3. Discovery | Asks how the buyer handles the problem today and what it costs them | Notes in the buyer's own words |
| 4. Demo and next step | Shows the outcome that matters to this buyer and asks for a concrete next step | Next steps agreed or declined, and why |
| 5. Update | Records objections, wins and losses, and changes the hypothesis | A revised ICP and pitch, and an objection log |

The founder owns this loop because the learning is the point. Objections reach the person who can change the product, the price or the pitch the same day, and the pattern a first sales hire needs only exists once someone has written it down. Alex Kracov, CEO and co-founder of Dock, describes [his version](https://www.kracov.co/writing/founder-sales): he positioned each early call as a "feedback call", found many of them through introductions from investors and friends and through his own LinkedIn posts, and writes: "After every call, you need to look and think about how you can evolve the pitch."

For deciding which leads are worth a conversation, see our [lead qualification checklist](/guides/lead-qualification); for preparing the call itself, [Claude for sales](/guides/claude-for-sales). The AI agent sections further down help with step 2 – the outreach – not with the learning.

## How long should a founder lead sales – and when to hire the first sales rep?

There is no single answer, and the practitioners set the trigger differently:

- **Kazanjy, as [Dock](https://www.dock.us/library/founder-led-sales) reports it,** "doesn't recommend transitioning until you have done at least 50 sales demos and have a repeatable win rate of 20% or higher".
- **Jason Lemkin** of [SaaStr](https://www.saastr.com/the-founders-guide-to-transitioning-from-founder-led-sales-why-most-get-it-wrong-and-how-to-get-it-right/) puts the start of scaling "typically around $1M-$2M ARR" and says to bring on account executives first rather than a VP of Sales – then to get two AEs hitting quota before hiring a head of sales, around $2M–$3M ARR for sales-led companies.
- **Seth DeHart** ties it to the process rather than a number: hire only once you have a "(roughly) repeatable sales process", and test it by asking: "Could you document the whole process and let someone else run it without your involvement?"
- **On who to hire,** the GTMnow write-up of Kazanjy's workshop says: "Hire to execute the motion, not to discover it." The right first hire is "a high-ceiling, somewhat experienced seller".

The common thread: hire when the motion exists and is written down, not when you are tired of selling. A rep brought in to discover the motion inherits the founder's hardest job without the founder's authority to change the product. And the founder does not leave sales entirely – as Lemkin puts it: "The goal isn't to remove yourself from sales entirely."

## Where a founder's week goes on LinkedIn

For B2B founders, LinkedIn is often where step 2 of the loop happens – the outreach to people who match the hypothesis. Most of that work is legwork an AI agent can do on your own account, and some of it is judgement only you can make:

| Step | What an AI agent can do | What stays with the founder |
| --- | --- | --- |
| Build the list | Run your saved LinkedIn or Sales Navigator search, or search by title, industry and location | Setting the ICP |
| Research | Read each person's profile, current role and recent posts | – |
| Avoid duplicates | Check whether you are already connected or have an invitation pending | – |
| Write the note | Draft a short note that refers to something the person wrote | Approving every note |
| Send | Send the approved invitations at a human pace, inside your account's limits | – |
| Accepts and replies | Report who accepted and who replied | The conversation |
| Follow up | Draft follow-ups after checking the inbox | Approving every follow-up |
| Post | Publish or comment from a draft you approved | The voice and the opinion |
| Calls, demos, pricing, closing | – | All of it |

The agent works through [Linked API](/) on your own LinkedIn account, in a real cloud browser at a human pace – Linked API never stores your password. Connect it once and ask in plain language: as a custom connector in Claude (no code), through the [MCP server](/mcp/installation) in any MCP client, or with the agent-friendly [CLI](/cli/getting-started) in Claude Code or Codex. How agents connect to a LinkedIn account in general is covered in the [LinkedIn AI agent guide](/guides/linkedin-ai-agent).

## A worked run: Monday's LinkedIn outbound with an AI agent

Here is one week's outreach step, from a saved search to approved notes, plus a look back at last week. Both examples assume [network monitoring](/docs/monitoring-network) was on before the week they report on – switching it on collects new acceptances and does not backfill earlier ones. The first example also reads the inbox feed, which holds only messages that arrived after [inbox monitoring](/docs/monitoring-inbox) was enabled; before any follow-up, the agent reads each person's own conversation instead, so an empty feed never counts as "no reply".

**You:** Here's my saved Sales Navigator search and my ICP: heads of operations at logistics software companies in the US with 50–500 employees, who talk about manual processes. Find 20 people I'm not connected to yet, read their posts and comments from the last 30 days, and draft a connection note under 200 characters that refers to something they wrote – SKIP anyone with nothing specific. Then tell me who accepted last week's invitations and who replied. Send nothing until I approve.

**Agent** (used Linked API: nv_search_people, check_connection_status ×20, fetch_person ×17, get_network, get_inbox):
- 20 leads from the saved search URL
- 3 already connected or with an invitation pending
- current role, posts and comments from the last 30 days
- invitations accepted in the last 7 days
- messages received since last Monday

14 notes drafted and 3 people skipped with nothing specific to refer to. From last week: 4 people accepted and 2 replied. Nothing has been sent. Three of the 17:

| Person | Why they fit | What they wrote | Draft note |
| --- | --- | --- | --- |
| Rachel T. | Head of Operations, 140-person freight software company | A post on moving carrier onboarding off spreadsheets | "Hi Rachel – your post on getting carrier onboarding out of spreadsheets hit home. Would be glad to connect." |
| Marcus L. | VP Operations, route-planning software, 300 employees | A comment on what dispatch automation still gets wrong | "Hi Marcus – your comment on what dispatch automation still misses was sharp. Happy to connect." |
| Priya D. | Head of Ops, warehouse software startup | Nothing in the last 30 days | SKIP – nothing specific to refer to |

*Nothing goes out until you approve; then the agent sends the approved notes at a human pace on your own account.*

*Example run. Sales Navigator search needs a Sales Navigator seat on the account and the Plus plan; a saved standard LinkedIn search works the same way. Assumes network and inbox monitoring were on all of last week – they collect new activity only. Works in Claude, ChatGPT, Cursor or Claude Code through the MCP server or the CLI.*

Why approve every note: it goes out under your name, to people you may sell to for years. LinkedIn allows 200 characters in a note on a free account and 300 on Premium, and it limits how many personalized notes an account can send – when that limit hits, the invitation can still go out without a note. What to write, and when to skip the note entirely, is in [LinkedIn connection request message examples](/guides/linkedin-connection-request-message); how many invitations are sensible per week is in the [connection limit guide](/guides/linkedin-connection-limit-2026). For a deeper version of the find-and-qualify step, see [AI for sales prospecting on LinkedIn](/guides/ai-sales-prospecting-linkedin).

## Follow-ups and replies without a sequencer

Linked API has no sequencer and no automatic stop when someone replies. Each follow-up is a new run you approve, and the inbox check comes first:

**You:** Draft a short follow-up for everyone who accepted my invitation in the last 7 days and hasn't written to me. Check the inbox first and skip anyone who already replied. Show me the drafts – don't send.

**Agent** (used Linked API: get_network, sync_conversation ×6, get_inbox ×6, fetch_person ×4):
- 6 invitations accepted in the last 7 days
- each person's conversation refreshed from LinkedIn
- each person's conversation read; 2 of the 6 already replied – left for you
- current role and latest posts

4 follow-ups drafted. The 2 people who replied are waiting for you in the inbox. Nothing has been sent. Two of the drafts:

| Person | Accepted | Draft follow-up |
| --- | --- | --- |
| Rachel T. | Tuesday | "Thanks for connecting, Rachel. Is carrier onboarding still the most manual step for your team, or has that moved on?" |
| Dan K. | Thursday | "Thanks for connecting, Dan. Saw you're expanding the ops team – happy to compare notes on what you're automating first." |

*The agent sends each approved follow-up as a single message; it does not schedule the next one.*

*Example run. Assumes network monitoring was on for the whole 7 days – it collects new acceptances only. Replies are checked in each person's own conversation. Works in Claude, ChatGPT, Cursor or Claude Code through the MCP server or the CLI.*

Messages go only to people you are connected with. If you would rather be pinged than ask, Linked API can send an `inbox.messageReceived` webhook to your own tools when someone writes, and a `network.connectionAccepted` event when an invitation is accepted ([webhooks](/docs/webhooks)). Cadence and message templates are in the [cold outreach guide](/guides/linkedin-cold-outreach-strategies-2026); what to do once a reply turns into a call is in [Claude for sales](/guides/claude-for-sales).

## Posting and warm signals

Outbound lands better with people who already recognise your name. A founder's own posts and comments build that, and the voice has to stay the founder's – an agent can publish a post or a comment from a draft you approved, not form your opinions for you.

Warm leads are an on-demand pull today: ask the agent who viewed your profile, or who reacted to and commented on a post in your space, and add the ones who fit to Monday's list. Our guides to [who viewed your LinkedIn profile](/guides/linkedin-who-viewed-my-profile) and [LinkedIn buying signals](/guides/linkedin-buying-signals) cover both.

## Handing outbound to your first sales hire

When you hire, hand over the motion, not just the contacts. The GTMnow write-up of Kazanjy's workshop lists what should exist in writing first: a written ICP, a prospecting playbook, a discovery guide, a demo script, an objection guide, a competitive sheet and defined deal stages.

If an agent has been running your outreach, part of the prospecting playbook already exists as artifacts:

- **The saved search URL** – the list definition, exactly as you ran it.
- **The ICP and the prompt** – the rules the agent applied, in plain language.
- **The notes and follow-ups that got replies** – your real message-market fit, not a template.
- **The reply log** – who answered, what they said and what happened next.

The new hire connects their own LinkedIn account – one seat per account – and runs the same prompts under their own name. They start with your evidence instead of a blank page, and you can check what GTMnow's write-up calls the critical success metric – "whether the rep's conversion metrics track close to yours" – from the first week.

## What outbound – and an agent – can't fix

- **Positioning and ICP.** If the message is wrong for the market, more invitations make it fail faster. That is the job of the learning loop above.
- **Contact data.** Linked API returns LinkedIn data, not personal emails or phone numbers, and it sends no email.
- **Sequences and booking.** No multi-step sequencer, no automatic stop on reply, no meeting booking, and no CRM write-back unless you build it, for example with [n8n](/integrations/n8n/overview) or [Make](/integrations/make/overview).
- **LinkedIn's own limits.** Invitation and note limits still apply, and the pace stays human.
- **The selling.** The agent finds, reads, drafts and reports. Discovery, demos, pricing and closing are yours – that is what founder-led means.

If you would rather buy a packaged AI SDR than assemble your own, our guide to [AI BDRs](/guides/ai-bdr) walks through that choice.

## Frequently Asked Questions (FAQ)

#### What is founder-led sales?

Founder-led sales is the stage when a startup's founder sells personally – finding the first customers, running discovery and closing – and works out the messaging and the ideal customer profile before hiring sales reps. Pete Kazanjy's *Founding Sales* describes it as discovering, refining and scaling the initial sales motion, then handing it to professional sales leadership.

#### How do I do founder-led sales on LinkedIn without a sales team?

Start from a written ICP and a saved LinkedIn or Sales Navigator search. Each week, have an AI agent build the list, read profiles and recent posts, and draft short notes that refer to something the person wrote; approve each note before it is sent. Check who accepted and who replied, and keep the conversations yourself.

#### How long should a founder lead sales?

Until the motion is repeatable and written down. Practitioners give different triggers: Kazanjy, as Dock reports it, at least 50 demos and a repeatable win rate of 20% or higher; Jason Lemkin, first account executives once you start scaling, typically around $1M–$2M in annual recurring revenue. Most add that a founder never leaves sales entirely.

#### When should a founder hire the first sales rep?

When the sales process is documented and repeatable – Seth DeHart's test is whether someone else could run it from your documentation without you. Hire someone to execute a motion that already works, not to discover one.

#### What should I hand over to my first sales hire?

A written ICP, a prospecting playbook, a discovery guide, a demo script, an objection guide, a competitive sheet and defined deal stages – plus, if an agent ran your outreach, the saved search, the prompt, the notes that got replies and the reply log.

#### Can an AI agent run founder-led sales for me?

It can run the outreach legwork – lists, research, draft notes, acceptances and replies – on your own LinkedIn account, waiting for your approval before anything is sent. It cannot do the selling: discovery, demos, pricing and closing stay with the founder.

#### Is founder-led sales the same as founder-led growth or founder-led marketing?

No. Founder-led sales is about who sells – the founder, until the motion is handed to a sales team. Founder-led growth and founder-led marketing are about the founder's own audience and content as a growth channel.

## Hand the legwork to an agent, keep the selling

Get it out of the box: connect the [MCP server](/mcp/installation), the agent-friendly [CLI](/cli/getting-started) or a ready-made [skill](/skills) and run the Monday routine from a plain-language ask. Or build it into your own tools with the REST API, the Node and Python [SDKs](/sdks/installation) and the shell CLI. Either way it runs on your own LinkedIn account. Plans are flat per seat, from $49/mo billed annually – see [pricing](/pricing).

*Facts verified November 4, 2026 – quotes and figures checked against Founding Sales, Alex Kracov's essay, GTMnow, Dock, SaaStr and Growth Unhinged on that date.*