An AI Agent That Writes Upwork Proposals Hit $10K MRR in 60 Days
After exiting a 15-person dev agency, Ivan Nedelkovski built Lancer, an AI agent that finds jobs and writes proposals on Upwork, and reached $10K MRR within 60 days of launch
Continue exploring this workflow
The Strategy
Ivan Nedelkovski built Lancer after already having run and exited a 15-person development agency, which gives this story a different starting point than most solo-builder narratives in this directory: he understood the freelance client-acquisition problem intimately before writing a line of code, because sourcing and winning work on Upwork was a pain point his own agency had lived through repeatedly. The product automates the two most time-consuming parts of freelancing on Upwork: discovering relevant job postings as they come in, and writing tailored, high-quality proposals fast enough to be among the first responses a client sees, which matters significantly for win rate on the platform. Nedelkovski built a weekend minimum viable product back in 2024, then spent roughly six months with a two-person team turning it into a commercial v1.0 product, routing LLM calls through OpenRouter to manage model costs and reliability across providers. The path from prototype to paying product took real iteration time, which is a useful counterweight to the many stories in this directory that emphasize speed above all else. Six months between a working weekend MVP and a sellable commercial product reflects the actual gap between something that works for one person and something reliable enough to sell to strangers. Once launched commercially, growth was fast: $10,000 in monthly recurring revenue within about 60 days, and the product has since grown further to over $20,000 MRR with around 100 paying users at an average revenue per user of roughly $300 a month, a healthy price point for a tool aimed at freelancers who are themselves trying to win higher-value client work.
How It Works
Draw on direct prior experience with the problem being solved, in this case running an agency that depended on winning Upwork proposals, to design a product that addresses the real pain point rather than a guessed one.
Build a fast weekend prototype first to validate the core mechanic: automated job discovery paired with automated proposal generation.
Spend the following months, with a small team, hardening the prototype into a reliable commercial product rather than shipping the MVP directly to paying customers.
Route LLM calls through OpenRouter to manage cost and reliability across multiple model providers rather than depending on a single API.
Focus the product specifically on speed of response, since being among the first proposals a client sees on Upwork materially affects win rate.
Price the product to reflect the value freelancers get from winning more and better client work, landing around $300 a month average revenue per user.
Launch commercially and track monthly recurring revenue growth closely in the weeks immediately following launch.
Results
Ivan Nedelkovski reports reaching $10,000 in monthly recurring revenue within roughly 60 days of Lancer's commercial launch, since grown to over $20,000 MRR with approximately 100 paying users at an average revenue per user of about $300 a month. These are self-reported figures published on Indie Hackers and have not been independently verified.
Our Take
The prior-agency-exit context matters here more than in most stories in this directory: Nedelkovski was not guessing at a freelancer pain point, he had directly experienced it running a 15-person team that depended on winning Upwork work. That kind of founder-market fit tends to produce sharper products than a purely speculative build. The honest six-month gap between weekend prototype and commercial v1.0 is also a useful corrective to the instant-launch narratives elsewhere in this space; automating proposal writing well enough to actually win more work, rather than just producing more proposals faster, took real iteration. At roughly $300 average revenue per user, this is a premium price for a freelancer tool, which suggests real perceived value rather than just cheap volume pricing. Best suited as a case study for builders targeting other freelancers and solo operators as customers.
Frequently Asked Questions
The practical questions a builder or operator is likely to ask before trying a strategy like this.
What does this freelancing AI sales agent actually do?
This freelancing AI sales agent is a real workflow where the agent takes on an operational job, not just a brainstorming task. An AI Agent That Writes Upwork Proposals Hit $10K MRR in 60 Days shows what that looks like in practice. After exiting a 15-person dev agency, Ivan Nedelkovski built Lancer, an AI agent that finds jobs and writes proposals on Upwork, and reached $10K MRR within 60 days of launch The practical value comes from the agent handling repeatable business work with enough autonomy that a human only steps in after context has already been gathered.
Who should use a freelancing AI sales agent like this?
This example is most relevant for freelancing operators. It is especially relevant for businesses where speed to lead, after-hours coverage, or consistent intake quality directly affects revenue. The category here is Marketing & Sales, which means the best fit is a team looking to turn a manual bottleneck into a repeatable system with a freelancing AI sales agent.
Which tools are used in this freelancing AI sales agent setup?
The source names OpenRouter, Upwork. That matters because one of the strongest signals in this directory is whether the operator shared the actual stack. Named tools make a freelancing AI sales agent strategy far more useful than vague claims about “an AI system” doing the work.
How hard is it to implement a freelancing AI sales agent like this?
Intermediate difficulty is the current read. The listing suggests a launch window of months. Startup cost is listed as $50-200/mo. We were able to extract 7 concrete workflow steps from the source. We would treat a freelancing AI sales agent like this as a workflow that needs real business context, testing, and exception handling rather than something you should copy blindly from one prompt.
What results can a freelancing AI sales agent produce?
Ivan Nedelkovski reports reaching $10,000 in monthly recurring revenue within roughly 60 days of Lancer's commercial launch, since grown to over $20,000 MRR with approximately 100 paying users at an average revenue per user of about $300 a month. These are self-reported figures published on Indie Hackers and have not been independently verified.
How credible is this freelancing AI sales agent case study?
Right now the evidence comes from an Indie Hackers post. That is enough for us to study and curate the workflow, but not enough on its own to treat this freelancing AI sales agent like an audited case study. We look for named tools, concrete results, and enough workflow detail to understand what was actually deployed, then we add our own editorial judgment on top.
Related Strategies
More AI agent strategies you might find useful
Fully Autonomous Meta Ads Manager Built on OpenClaw
A fully autonomous Meta ads operation running on OpenClaw for $0 per month that …
SaaStr Replaced a 10 Person Sales Team With 20 AI Agents and 1.2 Humans
A sales team of 10 SDRs and AEs was replaced by 20 AI agents managed by 1.2 huma…
An AI Marketing Product That Grew to $30K MRR by Dominating Reddit
An AI marketing tool reached $30K MRR by building distribution entirely on Reddi…
A Complete Cold Email Campaign Built in 15 Minutes With Claude Code
From raw company list to a live campaign draft in 15 minutes using Claude Code, …
An n8n Workflow With 47 Nodes That Automates Freelance Bidding and 3x Revenue
A 47 node n8n workflow that finds freelance projects, analyzes them with AI, gen…
The AI Solopreneur Stack: Running an Agency with Zero Employees Using Autonomous Agents
A complete AI tech stack that replaces traditional agency overhead with autonomo…
Want more strategies like this?
Get weekly AI agent case studies in your inbox.