Getting your first ten customers for an AI product

The founders who get their first ten customers fast have one thing in common: they do not wait for customers to find them.

This sounds obvious. In practice, almost every first-time founder spends the first month after shipping building a waitlist, sending cold emails into the void, and wondering why nobody is signing up. The problem is not the product. The problem is the distribution strategy, which at this stage is not a strategy — it is a hope.

Ten customers is not a lot. It is also not automatic. Here is what actually works, and in what order.

Start with who you know

Your first customer is almost certainly someone you already know, or someone one degree away from you. This is not networking — it is the reality that trust is the primary barrier to a new product, and trust is easiest to transfer when there is already a relationship.

Make a list of every person you know who:

• Has the problem your product solves
• Trusts you enough to try something that is not polished yet
• Will tell you honestly when it does not work

This list is probably longer than you think. Work it before you do anything else. The goal is not a paying customer yet — it is a user who gives you real feedback. Payment comes after you have earned it.

How to ask: Be direct and specific. "I built something I think would save you time on X. I am looking for five people to try it for free for thirty days and tell me what's broken. Would you be one of them?"

Do not ask if they are interested. Ask if they will commit. Interested gets you a vague yes and no follow-through. Commit gets you a calendar invite.

Go where your customers already are

After the personal network, the fastest path to early customers is a community they already belong to — not one you have to build.

Communities for most B2B personas already exist: Slack groups for CS professionals, subreddits for founders, LinkedIn communities for ops leaders, Discord servers for product managers. Your customers are already talking to each other about their problems. Your job is to find where they are and show up there with genuine value before asking for anything.

What genuine value looks like: answering questions, sharing something useful, making an observation that only someone who understands the problem well could make. Not posting about your product.

After you have contributed enough to be a recognizable name — which takes weeks, not days — you can mention what you built in context. Not as an ad. As "I built this because I was frustrated with the same thing you're describing." The distinction matters.

Direct outbound, done specifically

Cold outreach works when it is specific enough to not feel cold.

The generic version does not work: "I see you work in CS. We built a tool that saves CS teams time." Delete.

The specific version does work: "I read your post in [community] about QBR prep. We built something that cuts the data-gathering part from four hours to forty minutes — would you be willing to try it for one QBR and tell me if that's actually true for you?"

The specific version works because it proves you know something about this person's actual situation, it makes a concrete and testable claim, and it asks for something small (one QBR, not a subscription).

At this stage, send twenty highly specific emails rather than two hundred generic ones. The response rate will be ten times higher and the feedback will be ten times more useful.

Finding contacts: LinkedIn, community member lists, conference speaker lists, Twitter/X bios, newsletter subscriber profiles when they are public. You are looking for the specific person at the specific company with the specific role — not "CS team at SaaS companies."

Use the product to get customers

The fastest way to demonstrate that your product works is to use it in the process of getting customers.

If your product generates something — reports, analyses, summaries, copy — use it to create something of value for a prospective customer before they even sign up. Send them a sample output made from their public data. Show them what their QBR would look like if you prepped it. Generate a competitive analysis for their market using your tool and share it.

This does the work of a demo without requiring a meeting. It shows, not tells. And it creates a moment of "how did they do that?" which is the best conversation opener there is.

The category of marketing this falls into is sometimes called "show don't tell" or "productize the pitch." It works especially well for AI products because AI outputs are immediately tangible in a way that a product roadmap or feature list is not.

The customer who refers you

Ten customers is actually a smaller number than it sounds, because you only need to find three or four of them yourself. The rest come from referrals.

The referral trigger: a customer who got real value and knows someone else who would get the same value. Referrals do not happen automatically — you have to ask. "Is there anyone else you work with who has the same headache around QBR prep? Would you be willing to introduce me?"

The timing of the ask matters. Do not ask before they have gotten value — they will not refer you to someone they respect. Ask after the specific moment when they saw something that impressed them.

Build the referral ask into your process, not as an afterthought. At the thirty-day mark, have a conversation with every early user. Ask what is working, what is not, and whether there is someone they would refer.

Partnerships with people who already have your customer

If there is a tool, community, or service your target customer already uses and trusts, the person running that tool is potentially your best distribution channel.

What this looks like in practice: your customers use a specific CRM. You build a native integration with that CRM and reach out to the CRM's partnership team. Your customers read a specific newsletter. You pitch the newsletter author on a mention or a co-promotion. Your customers use a specific agency for a task adjacent to yours. You offer the agency a referral arrangement.

This is a slower path than direct outreach, but it scales in a way direct outreach does not. One good partnership can be worth a hundred cold emails over time.

At ten customers you are too early for formal partnerships. But you can start building the relationships that lead to them.

What not to do

Ads. Not yet. Paid acquisition before you have validated the product-customer fit means you are paying to bring in customers who will churn, and using the churn as feedback when the problem might be acquisition targeting, not the product. Get to twenty customers organically first.

Content marketing. Also not yet — not as a primary channel. Content takes months to compound. You need customers faster than that. Write content once you have customers and know what they care about. Not before.

Building features to compensate for lack of customers. This is the trap. The product is not the blocker. Your next customer is probably one conversation away. Go have the conversation.

The honest number

At the earliest stage — before you have brand, before you have case studies, before anyone has heard of you — getting from zero to ten customers takes somewhere between two weeks and three months. The variance is almost entirely about how systematically and personally you pursue it, not about the quality of the product.

The founders who move fastest do the thing that feels uncomfortable: they reach out directly, they follow up when they do not hear back, they ask for referrals explicitly, and they have the hard conversation with users who stopped engaging instead of waiting to see if they come back.

Ten customers is not scale. It is proof. And proof is what everything else — funding, partnerships, growth — runs on.

Further reading

• How three YC startups built their companies with Claude Code — real founder go-to-market stories