Ivan spent 5 years running a 7-figure dev agency, sourcing all clients from Upwork. Qualifying jobs and writing proposals ate 10+ hours a week — perfect repetitive work for an AI agent. He coded the first version over a weekend using TypeScript, Next.js, and LLMs (for job qualification and proposal writing). It crushed internally, so he invited 2 agency-owner friends to beta test. They closed 3 five-figure clients in 2 weeks, proving it worked beyond his own agency.

• Identify a repetitive, high-value task you do weekly (job boards, lead research, outreach) that an LLM could automate
• Build a rough internal tool in a weekend using Next.js + OpenRouter APIs — optimize for speed, not polish
• Beta test with 2–3 friends in the same niche; if they close deals or save serious time in 2 weeks, you have product-market fit

Ivan launched Lancer with 3 pricing tiers: pay-as-you-go at $79 for 30 proposals, a light plan at $300 for 250 proposals, and an unlimited plan at $500/month (the original launch offer). Premium pricing for a SaaS, but justified by the 10+ hours saved and the five- to six-figure monthly revenue Upwork generates for agencies. The product hit $10,000 MRR in the third or fourth month after launch — no ads, no audience, just a sharp wedge into a platform with 200,000 jobs posted monthly.

• Price based on the value created (hours saved × hourly rate, or a % of revenue unlocked), not competitor benchmarks
• Offer tiered plans with usage limits to capture solopreneurs and agencies — anchor high with an 'unlimited' tier
• Launch fast once beta users prove they'll pay; you can always adjust pricing after real customer feedback

Instead of running ads or cold emailing thousands of freelancers, Ivan identified 'connectors' — Upwork coaches who already had 5–20 paying clients each month, charged $600–$1,000+ for coaching, and commanded massive trust. He closed the first coach through a beta user intro; the coach was blown away and started referring every client. The second coach he cold-messaged on LinkedIn, offering $1,000 just to demo the product and become a paid user. Both became affiliates: 30% lifetime commission if they onboard the client, 20% for referrals. These 2 coaches generated most of Lancer's growth — zero paid ads.

• Map your ICP, then find the 'layer above' — coaches, consultants, influencers who already have your customers and their trust
• Reach out with extreme personalization: reference their work, offer upfront payment ($500–$1K) to demo, or get a warm intro from a mutual connection
• Offer 20–30% lifetime commission (higher if they onboard), use Tolt or similar to automate tracking and monthly payouts

Once the first 2 Upwork coaches were onboarded, Ivan needed a system to track referrals and automate payouts without manual spreadsheets. He used Tolt, an affiliate marketing software, to handle attribution, commission splits (20% for referrals, 30% if the coach fully onboards the client), and monthly payouts. This let him scale the affiliate model without hiring a finance person — inbound Stripe sales just flowed in, and Tolt handled the rest. The key was making it dead simple for connectors to get paid, so they kept referring.

• Use Tolt, Rewardful, or PartnerStack to automate affiliate tracking and payouts from day one — don't build this yourself
• Set clear commission tiers (e.g., 20% referral, 30% full onboard) and pay monthly so affiliates stay motivated
• Make the affiliate dashboard transparent so partners can see their earnings in real time — trust = more referrals