← Back to Blog

    I Built a Lead Gen Tool in 2 Weeks That a Dev Shop Quoted $40K For

    ·Shubham Rasal

    A real case study of building a custom AI lead generation tool for an agency owner — timeline, stack, results, and what it actually cost.

    I Built a Lead Gen Tool in 2 Weeks That a Dev Shop Quoted $40K For

    The Problem Landed in My DMs at 11pm

    We found this client on Upwork and had this requirement for this tool. He wanted to automate his already proven workflow for lead gen that produces upwards of 20% reply rate which is unheard of.

    People like him are the best kind of people to work with since they are upfront with their requirements and know that there is a way to leverage software and ai to move faster.

    He was running a lead generation agency. His reply rates sat at around 20% — which is unheard of.

    The bottleneck wasn't his offer or his copywriting. It was the process that sat underneath everything.


    What His Workflow Actually Looked Like (Before)

    Before we touched anything, I asked him to walk me through exactly what his team did to process a single lead from start to first email sent. Here's what that looked like:

    1. They started with lead research which was done by humans
    2. Then manually update sheets and track all the leads
    3. It took days to do this with multiple people

    Total time per lead: 50 minutes. Multiply that by 5 days/week and you start to see why this was the ceiling of his business.


    What I Proposed (And What He Initially Thought It Would Cost)

    Before talking to me, he'd gotten a quote from a development agency. Their number was pretty high. Timeline: 3-4 months.

    I told him I could do it in 2 weeks.

    At first he was surprised as most agency owners don't know how to leverage AI to speed up their workflows.

    Here's what made that possible: I wasn't building from scratch. I was assembling.


    The Stack (And Why I Chose It)

    This is the part most "AI consultants" skip when they write case studies, so I'll be specific.

    Core tools:

    • Always use the latest models — Claude and GPT as of now
    • Cursor and Claude Code are the best harnesses for coding right now
    • Data was custom and proprietary
    • Custom CRM in a Postgres DB

    Why not just use Clay, Instantly, or Apollo?

    The client wanted a custom solution because something like that didn't exist out in the market. That's because he had a very unique workflow which was the reason he used to get such high reply rates and to build such a flow, we need custom tools to be used.


    How the Build Actually Went

    Week 1:

    The first week was pretty fast since we had to decide on the requirements and finalize on the scope of the project which was done within few days. And then we started on the development right away and came up with MVP pretty early so that we can get feedback from him and continue working on that. Since we have a full stack starter project template ready, we can get up and running with the database authentication and the full frontend and backend within hours so that there is no bottleneck on that side.

    Week 2:

    The second week was mainly for integrations and the AI features. That was because we needed a way to leverage AI to improve his workflows. We used it mainly for personalization of cold emails and researching the leads. Along with that, we had another feature for payment integration which was supposed to be like a credit system which we implemented right away.

    The moment it clicked:

    The real aha moment for the client was when we built the application so fast and he was shocked that we could do such great work in such less time. And all the features were functioning along with the integrations and he was pretty happy with that.


    The Results

    The tool enables him to track all the lead gatherers and sitters internally so that he has a complete overview of all the leads that are getting into the pipeline. He is able to integrate credit system because the plan is to sell this as a software to other agencies because this workflow is pretty profitable.

    The bigger play: he's now packaging this tool as a product to sell to other agencies in his space. That's the part that made this more than a client project — it's a business asset.


    Is a Custom AI Tool Right for Your Agency?

    Not always. Here's a quick filter:

    You probably need a custom build if:

    • Your workflow involves data from a proprietary or unusual source
    • Off-the-shelf tools require too many workarounds that break frequently
    • You want to own the tool as an asset (not pay SaaS fees forever)
    • You're planning to white-label or productize it

    You probably don't need one if:

    • An existing tool does 90% of what you need for $50/month
    • Your process changes too frequently to justify a build
    • You're not ready to invest in maintaining it over time

    Want the Same for Your Agency?

    I do a free 30-minute AI audit where we look at your current workflow and identify the single highest-leverage automation to build first.

    No pitch deck. No upsell. Just a map of where AI can actually help your specific business.

    Book the audit


    Shubham is the founder of Maximal Studio, an AI development agency that builds custom tools for agency owners. He's a software engineer with 6+ hackathon wins across web, mobile, and blockchain.