Hey Nick, I've created a sort of LM learning hub team that memorizes rag store data using simple prompts like 'store my names'. Nick, it gets saved to a database. So as time goes on, they can constantly store data and the bot can keep growing with their business and their team. Plethora of loose use cases for this ongoing store mechanic. I'm also using AI to create personalized emails. I'm going to use instantly or smartly to warm them and send. But I was wondering, I don't know if this will sell and me and my partner can create an automation. So, do you think a smart way of doing this is splitting emails? So, let's say we send 4500 emails a month, would it be smarter to send 1,500 for the custom LLM then, 1,500 for one automation and 1,500 for another or knuckle down on one? As I feel if I try and sell it, but it's not the best to sell/copy is bad, I might end up exhausting a lot of potential clients before we get what works. And for reading what I'm selling, do you personally think it can sell?
Yeah, for sure. But I think you went about this the wrong way. You started spending tons of time building this product. Now you're considering whether to sell this or something else. A lot of this is wasted time. You might have actually spent time selling a product that nobody cares about. Just to be blunt. The reality is, as beginners, we don't really know what we don't know. So the best way is to put as many possible things in front of as many people as possible. Put as many possible things in front of as many people as you can and let them guide you in the right direction. If you put five possible offers in front of 5,000 people and people disproportionately choose offer number three, then you can invest the time and energy necessary in building the product. This is the modern way of go to market: you go to market first, then build the product based on what people want. People respond positively to this. These are the questions I'm always getting about whether your automation does this or that. So that's what I'm going to build into the freaking thing. Does that make sense? So, it's putting the cart before the horse. Don't lose the forest for the trees. But that's my recommendation there.
The ceiling is virtually unlimited. The reason why is because cold email is quite productizable — there are only a couple of steps that really require a lot of human intellect and muscle. A quick example of how you could productize a cold email agency so you get almost unlimited output for the same fixed input: you could automate the process of buying domains, which is achievable via API endpoints. You can automate setting up mailboxes, also achievable via API endpoints or browser automation. You can automate finding offers that have worked really well for clients in similar verticals — presumably if you wanted to productize a cold email agency you'd just work with people in the same vertical, then copy and paste those offers, make minor context-dependent modifications, and run them. You can automate the process of replying, and you can automate iterating and managing that. Now you shouldn't automate everything, but when it comes to productization and scaling to 70K a month or more, you typically just need the same inputs done very similarly over and over again with minor modifications. Cold email agencies are probably one of the simplest and easiest ways to do so — that's a major chunk of the revenue that got me to over $70,000 in a month. So just productize the hell out of things across the stack. But note that when you start an automation agency you typically do custom work across a big swath of different client verticals and niches, and that's pretty difficult to scale to 70K a month. So typically what you have to do is close it down to a subset of those niches, sell a subset of the products you're currently selling to that subset of niches, and in that way make everything really similar over and over and over again.
Right now you're focused on the savings, and I think that's the main thing you've understood automations to provide, but savings is a small fraction of the total value of an automation. Say you build a system that saves 10 salespeople 1 hour a week each — that's 40 hours a month, and if they're paid $50/hour that looks like a $2,000/month automation. That looks sizable, but it's only a small part of the value you're capable of providing. The big thing isn't the savings, it's the revenue: if you free up an hour of a salesperson's time and they can use it to close deals worth, say, $500 in that hour (amortized across their whole workload), you're not just saving $2,000, you're making them an extra $20,000. That's the difference between a savings mindset and a revenue mindset — it's opportunity cost. When I sell a cold email system, I don't say 'this saves you $50/month on cheaper leads,' I say 'this frees up 10 hours a month of your salespeople's time, worth $500/hour, meaning it makes you $5,000.' Reorient your pitch around revenue, not savings — savings only matters if you're working with a multi-billion dollar enterprise. On your second issue: 3,000 emails to one booked meeting is a 0.03% booked-meeting rate. Two meetings would be 0.06%, three would be 0.1% — that's totally fine, not the best ever (ideally 0.5-1%, a meeting every ~200 emails), but it proves it's possible. You've run the equivalent of a 10-minute mile; now make your copy better and run a 9-minute mile.
Right now the offers that work are training/upskilling/workshops — getting businesses up and running with Claude Code specifically, including how to manage shared skills within it, since Claude Code is exciting to people but often outside their day-to-day skillset. For a meat-processing/distribution client specifically, we built an ERP-style inventory system with a lightweight AI layer that progressively takes over more of inventory management over time. You're right that the vast majority of my work these days isn't from cold traffic — I don't have a fixed set of offers right now, though I'd like to as I want to scale Left Click past $250K/month. Currently we're working with a few big names (e.g. Mr. Beast) and most of my time goes there, so it's mostly consultation. I also have team members now: Cody McLean handles a lot of the consultation/sales side (case studies with the government of California, Lululemon, etc.), and my business partner Geo leads the technical/dev side at Clear Vero. It would help my time efficiency to sell a simple productized package to small-to-mid businesses, but the current split leans toward larger clients for leverage reasons. We're actually spinning up a cold-traffic offer in the next few days, testing Instantly's new Airmail mailboxes through Left Click, to see how that performs.
If you are selling AI or automation services, working with local businesses is sort of a double-edged sword. The reason it's great is because you're working locally with businesses that typically don't have a lot of technical experience, so there's usually a big overhang between what is achievable with technology and what they are currently doing. Even a couple of little sprinkles of AI can really change their workflow and revolutionize their business. But the bad part of that double-edged sword is they're typically not very technically capable, which means they might struggle understanding the value of technology. You might actually have to build that value up in your explanations and sales calls from the floor all the way up to where the value actually is. A lot of people won't even realize why they might want to automate something like a purchase order automation — they'll just be like, well no, this is the same way we've always done things, and that's fine, and I don't really want to grow all that much anyway. That is pretty difficult to convince. If somebody's in a position where they're like, ah no, I'm cool doing it the same way I've always done it because they don't want more work anyway, how are you going to sell them on a big return on investment? Unfortunately, the same thing that leads to a lot of local business owners only having one simple brick-and-mortar storefront is the same thing that leads to them not really wanting to buy a service like this, because they just don't give a [care] about growth. Another thing is their data isn't scrapable, which is a problem. If you're trying to sell them voice and chat agents, it might be difficult to get enough of a base of local businesses in an email list to even pitch them a chat or voice agent to begin with. Your only way of doing that might be through something like cold calling, because a lot of these local businesses don't look at their email as much but they'll certainly look at their phone — they'll be around the front desk when people come into their shops. So that necessitates a change in strategy. These are all reasons why I don't really like local businesses. If you're wondering how you can implement AI features in digital businesses instead of local businesses, know that you are typically dealing with an audience that is more B2B oriented. They typically understand the economics of value exchange a little bit more. They understand the whole idea of, hey, I will provide you massive return on investment and I'll make a little bit for that, and here's how. They typically already use AI or have experience with AI, so it's not going to be starting from scratch — it'll just be a lot easier across the stack. And just be careful what you're selling them. For voice and chat agents for digital businesses, don't pick voice and chat agents before you've verified the niche. In general, don't pick the thing that you're going to sell before you know who you're going to sell it to. Experiment with a bunch of different approaches to a bunch of different people and then just go based off what works.
local businessdigital businessai agencyniche selection