How we built a custom AI cold email tool for a friend in two weeks

A friend running a small services business called us about a problem that did not sound like an automation project at first. He was spending close to six hours a week preparing cold outreach: pulling lists, copy-pasting into Mailchimp, rewriting subject lines for each segment, then bouncing over to Lemlist for follow-ups. Half his Sundays were going into this.

His real complaint was not the time. It was the quality. By the time he finished prepping a batch, his messages sounded like every other newsletter in the inbox. Generic openers, generic offers, generic results.

We considered telling him to stick with Lemlist or move to Apollo. But the founder had two specific needs that off-the-shelf tools handle poorly. First, he wanted personalization that actually read each recipient's website and recent activity, not just a {{first_name}} merge tag. Second, he wanted the whole thing to live inside his own SMTP and domain warmup setup, not behind a SaaS sender pool he did not control.

Most outreach platforms optimise for volume. He needed the opposite. Fewer emails, more thought per email, full control over deliverability. That is a different product.

Two weeks of focused work. The tool takes a CSV upload, enriches each row with light context about the recipient, and drafts a personalised email using a prompt template the operator controls. Nothing exotic on the AI side. The discipline was in the workflow around it.

Drafts go into a review queue before sending, so the founder can edit anything that feels off. Approved emails flow out through his own SMTP. Follow-ups trigger automatically based on opens, clicks, and replies. Bounces are tagged and quarantined. Replies thread back into a simple pipeline view so he can see who is warm, who is cold, and who is asking for a meeting.

We delivered it for $2K. No retainer, no per-seat pricing, no SaaS lock-in. He owns the deployment.

His outreach prep time dropped from around six hours a week to about thirty minutes. Most of that thirty minutes is now spent reviewing drafts and editing the few that need a human touch, which is exactly where his time should go.

In the first month he sent a few hundred emails, got a reply rate that was meaningfully higher than his Mailchimp baseline, and booked a handful of meetings. One or two of those turned into high-value conversations within the first two weeks. Real numbers will need to settle over a quarter, but the direction was clear.

Once the tool was stable on his side, we cloned the deployment for Sambhav Tech. We had been doing our own cold outreach the slow way: hand-written emails, spreadsheets, irregular follow-ups. Within a few weeks of using our own build, it brought us a paying healthcare client.

That was not the plan when we started the project. We kept eating our own cooking and it started feeding us back. The tool that was meant to save a friend six hours a week is now our default lead-gen engine.

If your outreach is mostly volume, the existing platforms are fine. If your outreach depends on context, judgment, and your own sender reputation, the off-the-shelf tools start to feel like a tax. You pay monthly for features you do not use and you fight against constraints you did not choose.

A custom build is not always the answer, but it is more accessible than most operators assume. Two weeks and a clear scope is enough to replace the pieces of your stack that are actively costing you time. The harder problem is usually deciding what not to build.

We now offer this as a custom build for $2K, scoped to your sender setup and your offer. We are also turning the same engine into a productized SaaS, releasing soon under the name Evos.

If you are spending more than four hours a week on cold outreach prep, talk to us at info@sambhavtech.in. Bring your current workflow and your last batch of replies. We will tell you honestly whether a custom tool, a productized version, or a better use of what you already own is the right answer.