Built Lancer, an AI agent that automates job discovery, qualification, and bidding on Upwork.
Previously ran NB Masters, a software development agency that did over seven figures and grew to almost 20 full-time employees.
Used Upwork as the agency’s main client acquisition channel and saw that it was an underrated source of quality leads.
Built the first version as an internal tool after noticing that qualifying jobs and writing personalized proposals took significant manual time.
Lancer
Helps freelancers and agencies turn Upwork into an automated acquisition channel.
Saves users more than 10 hours per week of repetitive work.
Uses LLMs for job qualification and proposal writing.
Offers a subscription model with pay-as-you-go, light, and unlimited plans.
Prices include $79 for 30 proposals, $300 for 250 proposals, and an initial $500 per month unlimited launch offer.
Metrics
Reached $10K per month in the third or fourth month after launching.
Helped beta users close three five-figure clients within two weeks.
Grew with zero paid ads.
Product origin
Started with a problem Ivan personally had inside his agency.
Saw that Upwork has around 200,000 jobs posted each month.
Treated the workflow as a strong AI-agent use case because jobs appear around the clock and each one takes time to qualify and answer manually.
Turned the internal tool into a standalone product after friends who owned agencies got results in beta.
AI agent opportunity
Sees strong potential in automating workflows that sit on top of existing software platforms.
Looks for platforms with large user bases and public customer activity because the users and pain points are easier to identify.
Uses Upwork as an example because ideal customers and their activity are visible on the marketplace.
Connector growth strategy
Identifies a layer above the ideal customer instead of selling directly through ads or cold email at scale.
Calls these people connectors because they already have trusted access to the target customers.
Chose Upwork coaches because they work with agencies and freelancers who want leads from Upwork.
Got most growth from two Upwork coaches, one through a beta-user intro and one through personalized LinkedIn outreach.
Offered commissions of 30% lifetime when a connector sells, onboards, and sets up a client, or 20% when they only refer.
Connector playbook
Define the ideal customer profile based on who onboards easily, pays, and stays.
Find people one level above that customer who have a trusted network of those buyers.
Send highly personalized pitches with research, references to their work, or a Loom video.
Work out incentives such as 20% to 30% lifetime commission or upfront payments when the connector’s network justifies it.
Track referrals and monthly payouts with affiliate software.
Tech stack
Built the internal weekend version before turning it into a commercial product.
Built the MVP over three months with TypeScript, Next.js, Node.js, GCP, and Firestore.
Uses Cursor with Opus, OpenRouter for LLM APIs, Hetzner and GCP for hosting, proxy providers for safe Upwork account connections, Elasticsearch for job data, and Tolt for affiliate marketing.
Advice
Start building software products sooner instead of delaying for income security.
Treat software product skills as highly leveraged because they apply across many online businesses.
Recognize that AI creates decades of opportunities to build useful products.
