I Built a $10K/Month AI Agent

Source
raw	raw/i-built-a-10k-month-ai-agent.md
url	I Built a $10K/Month AI Agent Ivan ran a highly profitable software development agency in Macedonia, NB Masters, which grew to 20 employees and over seven figures in revenue. Throughout five years of operation, their primary cl... https://www.youtube.com/watch?v=LKARRA0MvY4

TL;DR: Ivan, a software agency founder from Macedonia, built Lancer—an AI agent that automates qualifying and bidding on Upwork. He bypassed traditional marketing and ads, scaling the SaaS to $10,000 MRR in four months using a “Connector Marketing” strategy that turned Upwork career coaches into high-ticket affiliates.

The Founder’s Story

Ivan ran a highly profitable software development agency in Macedonia, NB Masters, which grew to 20 employees and over seven figures in revenue. Throughout five years of operation, their primary client acquisition channel was Upwork, a public talent marketplace posting over 200,000 jobs per month.

However, qualifying jobs and drafting highly personalized proposals to bypass spam filters is incredibly tedious, taking a human roughly 10 minutes per listing. Because Upwork jobs are posted around the clock, Ivan realized an AI agent could qualify and bid 10x faster. He built Lancer as an internal tool, which performed exceptionally well. After a small beta test with other agency owners yielded multiple five-figure clients within two weeks, Ivan launched Lancer as a standalone commercial SaaS. It hit $10,000 MRR in under four months.

The Building Process

Ivan built Lancer’s initial prototype over a single weekend for internal agency use. He then spent three months building the production MVP alongside his co-founder.

Vibe Coding Stack: Using Cursor with large models (including Claude 3.5 Sonnet / GPT-4 / Opus 4.5), the founders barely touched actual code, using natural language to build the backend logic.
Core Agent Logic: Lancer uses LLMs in two ways:
1. Job Qualification: Crawling Upwork jobs, checking parameters, filtering out low-quality requests or projects users want to avoid, and matching them against a comprehensive personal knowledge base.
2. Proposal Writing: Generating customized, context-aware cover letters matching the client’s needs, complete with required keywords and technical descriptions.
Safe Connection: Lancer routes actions through specialized proxy providers to connect securely and stably to users’ Upwork accounts without triggering security alerts.

The “Connector” Growth Strategy

Ivan used a B2B affiliate strategy he calls Connector Marketing. Instead of spending months building a social media audience or spending capital on cold emails and paid ads, Ivan targeted established figures who already possessed his Ideal Customer Profile’s (ICP) attention and trust.

Ivan’s 5-Step Connector Playbook

Strictly Define the High-Value ICP: Lancer’s pricing is premium for a SaaS ($79/mo pay-as-you-go, up to $500/mo unlimited). Ivan targeted high-volume agency users rather than casual freelancers because agencies send high volumes of bids, experience massive pain, and do not churn.
Locate the “Connector” Layer: Find professionals who run training programs, mastermind groups, or coaching services tailored to your ICP. For Lancer, these were Upwork career coaches who charge freelancers $600 to $1,000+ to help them get leads. These coaches already have massive inbound traffic, strong reputations, and high trust.
Draft a Highly Personalized Pitch: Do deep research on each coach. Pitch them via long Loom videos detailing how Lancer makes their student cohorts 10x more successful on Upwork. Ivan even made a cold outreach offer to a prominent coach, paying him $1,000 upfront just to jump on a call and test the tool.
Structure High-Margin Lifetime Commissions: Offer a compelling commission structure. If the coach actively onboards and sets up the customer on Lancer, they receive a 30% lifetime recurring commission. If they simply refer the user, they receive 20% lifetime recurring commission. Because Lancer is a high-priced B2B tool, this yields substantial monthly payouts for the coach. Most of Lancer’s growth came from just two coaches.
Automate Tracking & Payouts: Use professional affiliate marketing software like Tolt to track referrers across multiple connected dashboards and automate recurring monthly Stripe payouts.

The Tech Stack

Programming Environment: Next.js, Node.js, TypeScript (all written via Cursor with Opus 4.5 / Claude 3.5 Sonnet)
Hosting & Infrastructure: Hetzner, Google Cloud Platform (GCP)
Database & Search: Google Firestore, ElasticSearch (for high-speed job queries)
LLM Routing: OpenRouter API
Affiliate Tracking: Tolt (SaaS affiliate management software)
Accounts & Security: Multi-Proxy Providers (for safe client-side account integrations)

Key Quotes & Metrics

MRR: $10,000 MRR achieved by the 3rd or 4th month of launching.
Pricing Plans: $79/mo (pay-as-you-go / 30 proposals + $2 per extra bid), $300/mo (light / 250 proposals + $1.50 per extra bid), $500/mo (unlimited launch tier).
Ad Spend: $0 spent on paid advertising.
Agency vs. SaaS Leverage: “I delayed jumping into software products to optimize for financial security running a dev agency. Even though it made us a bunch of money, it was the wrong approach. The problems you solve building a software product are so highly leveraged. Mastering them is worth more than any six-figure salary.”
Core Opportunity: “Every platform that has a huge user base and tedious, manual work is something you can automate and build an AI agent on top of.”

micro-saas — Hyper-focused single-purpose SaaS and niche monopolies
no-audience-launch — Short-form UGC video algorithms and influencer profit-sharing
vibe-coding — AI-powered development and rapid MVP creation
idea-validation — Landing page buy buttons and paid intent testing