The ten-minute chore nobody wanted
Every Upwork freelancer knows the grind. Scroll through hundreds of listings, spend ten minutes qualifying each one, then write a customized proposal that probably goes unread. Repeat daily. It is repetitive, soul-crushing work, and it is exactly the kind of workflow large language models were made to absorb.
Ivan Nedelkovski saw the pattern clearly. The pain was universal among freelancers and agencies: qualifying jobs and personalizing proposals consumed ten or more hours a week, time that could have gone to billable work. The existing tools were either manual to-do lists or generic automation that missed the nuance of fit.
Lancer trains models to do the two things freelancers hate most. First, it evaluates job fit, scanning listings and scoring which ones actually match a freelancer's skills and rates. Second, it generates personalized cover letters for the qualified jobs, turning a ten-minute chore into something closer to a review-and-send step.
The product ships on a subscription model with tiered pricing — pay-as-you-go, light, and unlimited plans — so everyone from solo freelancers to agencies can scale their proposal volume cost-effectively. Simple, legible, and matched to how the customers already think about the work. Notably, the pricing mirrors the value directly: more proposals sent means more chances at revenue, so paying for volume feels like an investment rather than a cost. That alignment between price and perceived return is a quiet driver of conversion that many AI tools miss when they charge flat fees disconnected from outcomes.
Borrowed trust beats bought attention
Lancer reached $10,000 in monthly recurring revenue within a few months of launch. The striking part is not the number. It is the acquisition cost: zero dollars in paid advertising.
Ivan's distribution strategy had a name: connectors. Instead of buying attention, he borrowed trust. He identified the people his ideal customers already listened to — top Upwork coaches, the platform-native authorities who teach freelancers how to win clients — and recruited them as affiliates.
The economics were deliberately generous. Connectors earned 20 to 30 percent lifetime commissions on every customer they brought in. That is not a bounty; it is a partnership. One coach joined after a personal introduction through Ivan's network. Another was secured with a $1,000 LinkedIn outreach incentive, a small, targeted spend aimed at exactly one high-leverage relationship rather than sprayed across an ad platform.
The logic is worth unpacking. A coach's recommendation arrives pre-loaded with trust that no ad can buy. When someone who taught you Upwork says a tool is worth it, the evaluation is mostly done. Ivan essentially rented distribution from people who had already spent years building it.
Why the channel fit the product
The connector strategy worked because the product and the channel fit like a lock and key. Upwork coaches make their living helping freelancers win more work. A tool that automates proposal writing makes their students more successful, which makes the coaches look better. Recommending Lancer was not a favor to Ivan; it was a service to their own audiences.
This is the part most affiliate programs get wrong. They recruit anyone with traffic and wonder why conversion is thin. Ivan recruited a handful of people with the right traffic — the exact humans his ideal customers trusted on the exact topic his product solved. Fewer connectors, deeper alignment, better economics for everyone.
The lifetime commission structure deserves emphasis. A 20 to 30 percent cut forever is expensive on paper, and that is the point. It signals to the connector that this is a real partnership worth their reputation. Short-term bounties attract promoters who disappear after the payout. Lifetime revenue shares attract partners who stick around, give feedback, and keep recommending because their income grows with yours.
Within months, this network-based distribution delivered a steady stream of paying users, each arriving with higher intent and lower skepticism than any ad click could produce. There is a compounding effect worth noting too. Every coach who succeeded with the program became a proof point Ivan could show the next coach, making each subsequent recruitment easier than the last. Trust, once demonstrated, becomes its own sales collateral.
Boring tech, sharp distribution
Underneath the clever distribution sits a deliberately unglamorous build. Ivan shipped Lancer on TypeScript, Next.js, and Google Cloud Platform — the familiar stack he already knew. No exotic infrastructure, no resume-driven architecture, just shipping speed.
That choice reflects a broader lesson in the case. The scarce resource was not engineering novelty; it was distribution insight. Every week spent evaluating a new framework would have been a week not spent finding the next connector. Ivan put his innovation tokens into the go-to-market motion and kept the technology boring.
The bigger pattern is one founders should study. AI agents that layer onto existing marketplaces can unlock revenue streams that the marketplace itself leaves on the table. Upwork has millions of users and a well-known repetitive workflow. Lancer did not need to create demand or educate a market. It needed to remove one hated chore for people already spending money to solve adjacent problems.
Find a busy platform, identify its most repetitive workflow, define the humans everyone trusts there, and make them partners. That is the Lancer playbook, and it cost $1,000 plus generosity to execute.
The playbook
- Define your ICP with uncomfortable precision — Lancer serves Upwork freelancers and agencies. Ivan knew exactly who they were, where they hung out, and who they trusted before writing a line of distribution code.
- Recruit connectors, not influencers — Instead of ads, Ivan recruited top Upwork coaches — people with direct access to his ideal customers and pre-built trust. Authority transfers faster than it builds.
- Make the economics genuinely generous — Coaches earned 20-30% lifetime commissions. Generous, permanent economics align incentives; stingy one-time bounties attract mercenaries who churn.
- Build the MVP on boring, familiar tech — TypeScript, Next.js, GCP. Familiar tools meant shipping speed instead of yak-shaving on exotic infrastructure nobody would ever see.
- Layer AI on existing marketplaces — Upwork already had the users, the workflows, and the repetitive pain. Lancer just layered intelligence on top. Find a busy marketplace, then remove its most annoying chore.