Last month, I needed to land a handful of very specific SaaS founders for a new product launch. We’re talking founders who just raised a Series A, building in the dev tool space, and actively hiring for platform engineers. A needle in a haystack, right? My usual approach — buying a generic list and blasting out a slightly personalized template — was going to be a disaster. I knew it. I’ve been there, staring at dismal open rates and even worse reply rates, feeling like I’m just shouting into the void. That’s when I really dug into how AI-powered cold email strategies could actually help, not just in theory, but in practice, in 2026.
The promise of AI for sales is huge, but the reality often falls short. It’s easy to generate a thousand emails that sound like a robot wrote them, or worse, get yourself blacklisted because your personalization efforts are just plain creepy. What I needed wasn’t more volume; it was precision. I needed to find the right people, craft messages that resonated, and do it at a scale that didn’t require me to hire a small army of SDRs.
Finding the Right People (and Not Wasting Money)
My biggest gripe with cold email has always been the data. You can have the best message in the world, but if you’re sending it to the wrong person, it’s just noise. I’ve wasted thousands of dollars on lead lists that were outdated, inaccurate, or just plain irrelevant. You know the drill: you get a list of 10,000 ‘CTOs,’ only to find half of them are in completely unrelated industries or left their role three years ago. It’s infuriating.
This is where AI really starts to earn its keep. It’s not about magically generating leads from thin air, but about refining the data you already have or finding hyper-specific segments within larger databases. I’ve spent a lot of time comparing tools like Apollo and ZoomInfo. Honestly, for raw breadth and decent filtering, Apollo often wins out for me, especially if you’re comfortable getting your hands a little dirty with custom filters. ZoomInfo is powerful, sure, but its price tag feels a bit ridiculous for what you get if you’re not an enterprise shop needing every bell and whistle. For my specific founder search, I started with Apollo, pulling a broad list based on company size, industry, and recent funding events.
The AI component came in by feeding that initial list into a custom script using a tool like n8n for sales workflows or even a simple Python script with a bit of GPT-4o. I’d ask it to:
- Identify companies that specifically mentioned ‘dev tools’ or ‘platform engineering’ in their recent job postings or press releases.
- Filter out founders who had recently changed roles, indicating they might be less receptive to new pitches right now (or conversely, more open if the pitch was about their new venture).
- Cross-reference funding announcements with public tech stack data (available through various APIs) to ensure they were actually using relevant technologies.
This process took a few iterations, but it transformed a generic list of 5,000 potential founders into a hyper-targeted list of about 300. That’s a massive difference in quality, and it meant my outreach wasn’t just a shot in the dark.
Crafting Messages That Don’t Sound Like a Robot (Or a Sales Guy)
Once I had my golden list, the next challenge was writing emails that didn’t immediately scream ‘cold outreach.’ My concrete love here is the ability to generate genuinely insightful icebreakers that show you’ve done your homework. Forget the generic ‘I saw you work at X’ line. We’re past that.
I’d feed the AI (again, often a fine-tuned GPT model or even a specialized tool like Lindy.ai) specific data points for each lead:
- Their company’s recent funding round and the lead investor.
- A recent blog post they wrote or were featured in.
- A specific open-source project their company contributed to.
- A relevant quote from a podcast interview.
Then, I’d prompt the AI to generate a 1-2 sentence opening that referenced one of these points in a way that felt natural and not forced. The key isn’t to let the AI write the whole email, but to give it guardrails. I’d provide a template for the core value proposition and call to action, and then let it weave in the personalization. It’s about combining the AI’s pattern recognition with my human understanding of what makes a good sales email.
For sequencing and deliverability, I’ve played around with both Instantly and Lemlist. For my money, Instantly has been a more reliable workhorse, especially for scaling campaigns without blowing up your budget. It’s got the features you need for deliverability and A/B testing, and it doesn’t try to overcomplicate things with too many ‘AI magic’ features that often underperform anyway. Lemlist is good, but I think Instantly is a better value, particularly if you’re running multiple campaigns simultaneously. Their $97/mo plan for unlimited emails and multiple accounts is fair, especially when you consider the alternative costs of missed opportunities or hiring more staff.