Transcription
There's an AI tool that can build your entire M&A deal flow system. You might have heard of it, Cloud Code, but it is radically changing everything that we do inside of acquisition sphere, from coming up with deal sourcing ideas, pulling miss off markets, enriching the companies, even running ICP qualification on every single one of them, completely end to end.
So, in this video, I'm going to show you how we use Cloud Code to:
1. Share our buy box and screen in SOPs across the entire team through things called skills.
2. How we use it to plug into databases to build deal lists and enrich them very quickly.
3. A secret hack that we use for sub-agents to filter for our ICP and even draft the deal memo on every target.
4. And then a new kind of M&A research category that when we say Cloud Code is changing deal flow forever, that last thing is going to be the most important thing. So you want to stick around it, uh, 100%.
Now, let's talk about why Cloud Code is such an absolute game-changer. So, for starters, if you, if all you've ever used is just a chatbot, so like ChatGPT or Claude on the web, just the chat until this point. You ask a question, it gives you an answer back. Maybe it will search the web a little bit to find live information, but that's really it. For most people, Cloud Code is so amazing and it's different. It's part of the Claude, uh, platform, but it's so amazing because it's basically, break. Let's call that the sandbox of the chat window. Because when you enter something into a chat and you say, "Hey, look at the SIM on my computer." It's not connected to the files on your computer. It has no shot of being able to do that. Cloud Code, though, is connected to the files on your computer or the web. It can make tool calls. It can make API connection requests between two software. It can store data so that if you write your buy box once, you're always using that buy box going forward. Just a completely paradigm shift in comparison to what is just a regular chatbot.
Now, you might think this is really complicated to set up, but it's not the case. I've basically, I, I have some background in coding, but I don't need to touch the code at all right now, and it's super easy. All you have to do is go to claude.ai/code, and then you can click on terminal, you can click on desktop, you can download their web app. It's really whatever you prefer. And you probably don't want to use like things like VS Code or whatnot. I personally use the terminal directly. All you have to do is just copy and paste what I'm going to tell you into the terminal, and you're ready to go. Anytime you have a problem with the setup, what I do is just take a screenshot of it. I go to the chat version on the web and I just say, "Hey, tell me how do I fix this?" And it's always come out with the right case.
So now that you have, let's say, Cloud Code set up and you download things, let's talk about the first game-changing feature, which is called skills. Previously, before skills, people would train an AI to make it kind of own custom use cases. Might remember or heard about custom GPTs or kind of like understanding fine-tuning models. All of these things, they worked okay, but they weren't like amazing. All a skill is doing for us right now, it's literally a text file that you can plug into Cloud Code, and it can remember everything that you give it. So if you worked on this long workflow, let's say, or some kind of an SOP, if you want to find deals, you build some kind of a checklist or an SOP, you basically tell it, look, hey, first we do step one, then we do step two, then we do step three. Instead of telling it that again and again, every single time, you can just set the rules one time. Set the boundaries and the guidelines inside of those skills.
And so I think for M&A, there's two things that Cloud Code does so amazingly with these skills. One is you can make process yourself and work it basically back and forth with Cloud Code, eventually getting to the point where you say, "Great, this is exactly the way we want to screen every new teaser or find new deals every day or every week or every month based on the criteria." And the other thing is you can use other people's skills. Instead of putting in, let's say, all the time to figure out how an API works or what is an API or all the other edge cases, you can just download someone else's skills. You even have a, what we call a repository that I'm building up right now, calling ourselves like building M&A deal flow skills, which is pretty cool.
So right now, it's got four main skills. We have a deal screener skill that grades any teaser against our buy box based on every deal we've looked at in the last year. The way I trained it, I basically said something like, "Hey Claude, guess how good this deal would be based on the financials and the seller signals that you can find. Then check how it actually performed in diligence and improve your guessing." So it did that over more than a thousand iterations. So now it literally just guesses back and forth, and it knows what makes a deal worth pursuing. So we're not wasting time on every deal. If you're using, let's say, BizBuySell or scraping broker emails, there's little edge cases. So vocabulary in their listing formats that Claude is going to get confused with. That's all taken away. Now you can just plug in the skill right away here. And then if you want to, let's say, scrape sites like Acquire.com or pull broker teaser PDFs, we have a skill just for that. So you can quickly make offers and look at financial analyses of those deals. Plus some common deal flow lists. I have a database of every active SMB SaaS listings basically in the US. Every HVAC and home services business that hit size threshold, every rollup candidate in three verticals that I'm focused on. You can just point Cloud Code at this repository and get access to all of it.
So what I like to do is I, I use a software called WhisperFlow, and I can just talk to it. So I say something like, "Hey Claude, using the little screening skill, I just got a teaser for $2 million, a vertical SaaS in HVAC. Listing says 40% margin, 25% growth. Asking is 4x revenue, 12 employees, US-based. Score it against our buy criteria and tell me if it's worth a call." And so now we're just going to shoot that off to Cloud Code. It's going to read our skill document and then it's going to put that together. So you'll see it's even proposing the scoring breakdown for me. Hard criteria, soft criteria, risk flags. It's even asking me clarifying questions, which is great. I basically tell it, look, customer concentration matters more on this one. So flag if we don't have it disclosed or verified financials required, we can keep that. Thank you so much. And when you finish this, I tell him, "Open the scorecard so I can show everybody." Right now, we're just going to let that run. So, we get it scored 38 out of 50. And the recommendation is worth a screening call. It's flagged two risks: customer concentration unknown, churn data missing, without clicking around, without doing anything, just using my voice. And that deal screening skill. I now have a defensible scorecard on this teaser, really just in a flash, just like that, right? My favorite part about using my voice is then also going from one deal to 100 and cleaning up that data. That's use case two.
So now we got our scorecard without clicking around, just with my voice. But this is one deal. The real game is doing this for 100 deals at once and finding those hundreds of deals in the first place. And my favorite part about using my voice is then also building the list from scratch and cleaning up the data. So every week, our team is looking for at least 20 to 30 new targets matching our buy box. Before Cloud Code, this was like we needed like an analyst spending hours on BizBuySell and other marketplaces manually copying listings into spreadsheets. Now I just describe what I want to the AI. So I tell him, "Hey Claude, using the deal sourcing skill, build me a list of SaaS companies correctly, uh, for sale. ARR should be between 1 to 5 million. Uh, EBITDA margins above 25%, US-based, listed less than 30 days. Check BizBuySell and other marketplaces and our broker email digest folder. And I want output a CSV with the company name, asking price, revenue, EBITDA, industry, source, and days listed." So now we can get our CSV from multi-source scrape and without clicking around, without doing anything, just using my voice and that deal sourcing skill that we built. We now have a list of 23 active SMB SaaS listings in the US matching our exact buy box, just like that.
But the listing data is incomplete. The teaser doesn't tell us the website, the team size, and the tech stack or any recent signal about the seller. We need to enrich the list now, right? So here's where I'm going to show you a secret about how we're going to do that. We're going to switch back over to Cloud Code and I'm going to tell it this: "Thank for the list. Can you enrich each company? Find their website, LinkedIn, employee count, tech stack from job posting and any recent news. Use sub-agents for research on each company." So while that's thinking, the secret sauce that we've been using inside our acquisitions firm is this, right? Normally what you do have to do if you wanted to look at a company description is to ask, "Is this actually a SaaS company or a service business with a dashboard? Are they a good fit?" And you wanted to use AI to look at this, you'd say, "Hey, use our AI OpenAI key or whatever. Here's the prompt. Make this call." But Cloud Code gives you so much extra usage. On if you're under max plan, we are basically kind of like circumventing making direct API calls and just using set sub-agents inside Cloud Code to do all of our deal fit filtering, draft our deal memos, and that's exactly how we do this. I literally just say, "Prove to me that this is actually a SaaS company, not a service job," and use the sub-agents to do that. So now with all that, we learned a bit of the skills, how to find deals, how to look at a deal and filter them. So now let's jump back into the terminal screen and see how it's enriching all 23 companies. What I was talking about classifying batch one as SaaS, classifying batch two. This is where it's spinning up those sub-agents that I mentioned. So we don't have to pay extra. We don't have any extra token cost. It's not charging me any other way. I'm using my Cloud Code usage to get all of this done. Now we just have to wait. And when we're back and everything is running again, I didn't have to touch a single thing. Just with my voice, we enriched 23 companies, the websites, the headcount, the tech stack, recent news, and then the most interesting thing is we have the full company description. We have the output of, "Is it a software company? Is it a service shop in disguise?" And we have all the reasoning right here. So like I said, we use Sonnet, the sub-agents to be able to get this done. You could review it. You could get back and forth. You could say, "Actually, I prefer this is considered a service business, not SaaS." But that's one of my favorite hacks for our deal fit filtering.
So, the sub-agent thing, a lot of people don't know about it. By the way, quick pause, 'cause I know some of you watching this don't want to build any of this yourself. You want the deals, you want to offers made, you want to skip the part where just you want to close on deals and grow companies. And this is what we do at Acquisitions.com. We actually run this entire AI-powered system on your behalf. We find deals matching your criteria. We make the initial offers on your behalf. We analyze every deal that comes back to us. We connect you with banks. We structure financing. We run, we even connect you to investors, equity investors if you need a capital partner alongside the bank. By the time, uh, something hits your calendar, you basically just need to jump on calls where the deal is real, the seller is motivated, and the math works, right? With doing all the initial work. So, actually, here's actually where it gets unfair. After you close the deal, we can stay in with you. We can implement the same AI agents we built ourselves into your acquired company. It's the same process that helped me sell Wopups.com to Naval Ravikant's Group and the ones that help us grow Acquisitions.com revenue by 37% while we cut headcount by 80%. So it's AI for cost reduction, AI for revenue growth, AI for operations, plug in immediately before the deal and after the deal. Your job is just to show up to the meetings, like the deal, and sign. That's it. So, if you want, we're going to put a link at the bottom to book a call in the, um, description. We only work with a few people at a time, but anyway, looking forward to work with you if that's something you want to explore.
Now, back to use case number three, which is qualifying deals at scale. So, this is use case three, and it's genuinely my favorite hack. It's what we call kind of like sub-agent qualification. And here's a problem we try to solve. Let's say we have 23 listings here. Some of them look good on paper, but have hidden issues. The revenue, project-based. It's not really recurring. The SaaS is actually a services shop with a dashboard. Growth came from one whale customer that's about to churn. You need someone to help you read between the lines to see if it's a good deal or not. So, normally what we'd have to do is if we ever wanted to look at a company description and ask, basically look at it and say, "Hey, is this a good fit or a bad fit deal? Do they have customer concentration? Are they truly a software business?" And we wanted to use AI to do that. And we then say, "Hey, use our OpenAI key. Here's the prompt. Make this call." But Cloud Code right now gives you so much more. Like I said, with the plan, we're basically using the different API calls to help us with this. And I'll tell you what the prompt that we're using. And it's the same thing that in the past I had to pay like a junior analyst, spend two hours per deal. That's if 23 deals, that's like a full week of analyst time before we even like made a single call with a business owner or a broker. Now I just tell Cloud Code, this is the prompt for each company in our enriched list. "Run a qualification analysis using a set sub-agent. For each one, ask: Is the revenue truly recurring based on the employee count and revenue? Does it look like a software company or a service shop? Are there any customer concentration red flags? Is the multiple within our range? And what's the confidence score between 1 to 5 on each? And I want you to give me final recommendation: Do I need to pursue, pass, or I need more information?" It will probably check with me to make sure that what it thinks counts as recurring is actually what I mean by recurring before that runs. But that's the awesome power of using this through Cloud Code. This is going to be included in basically in my plan usage to be able to run it like this. And I don't have to pay anything extra.
So now let's jump back to the terminal. I'll see if spinning, how it's spinning sub-agents. It's qualifying batch one. It's qualifying the second batch and the third batch. This is where it's spinning up those agents. So we don't have to pay anything extra with those sub-agents. That's really cool if you're doing it on your own. So no extra token cost. It's not charging me in any other way. Then we wait, I guess. And when we're back, out of the 23 deals, it recommended pursuing eight, passing on 11, and flagged four as needed more info. Uh, the passes, three were service businesses, mislabeled as SaaS companies, exactly what we asked it to catch. Two had multiples way above our range, and the rest had, uh, basically deal-breaker, uh, red flags. 18 minutes total, zero extra API spend, and we have the reasoning sitting right here for every single one. You can review the deal, you can push back, you can override it. It's pretty, pretty cool.
So, last thing, in my opinion, is the biggest game-changing thing and thing, uh, we're working on the most right now for our customer, and it's something we call autonomous deal monitoring. I think this is by far the most game-changing part about Cloud Code with everything deal research. And the simple idea, I mean, when you give Cloud Code this thing, its ability to run this process over and over again autonomously is a game-changer. And we'll talk about that in a second. All this does is basically it's a framework where you say, "I'm going to give you a bunch of context about how my acquisition sphere worked, what our buy box, and we passed on, uh, in the past, and the experiments I want to run. I want it to optimize for one thing: the number of qualified deals that actually convert into closed acquisitions." Or if you're using this outside of M&A, you can even optimize it for whatever is your equivalent thing. I don't know, whatever outcome you want on your deals or leads. So then what you can do is you could use this framework to keep running experiments without you touching anything. It improves over time. You're sourcing different marketplaces or channels or, uh, screening criteria, different outreach copy, different offer terms, whatever it might be, without touching anything. You can have an AI optimizing your deal flow, running experiments, and then learning from what works and what doesn't work. So you don't have to keep running those experiments yourself because deal flow is just a fancy word for testing. Testing new channels, testing criteria, testing offers. And with this autonomous framework, you'd be able to just plug it in without you having to think about anything other than just giving it context front to basically say, "Hey, look, don't ever offer, let's say, about six times EBITDA on a deal that's growing another 20%." Here are some other rules of experiments you can run. You can say, "Hey, it can plug into BizBuySell, plug into broker email, plug into Affinity or whatever CRM you're using, change the buy box parameters, change the outreach to the broker, and run all these experiments for you on your own behalf." The prompts you can use for this whole thing is set up autonomous deal monitor. "Every day at 6 a.m., pull new listings from BizBuySell, Acquire, and other marketplaces. Uh, broker leads, reread the deal screening skills. Screen each new deal. Enrich the ones that pass with sub-agents. Drop a daily pipeline report and send, uh, to Slack a message. Track which deals to pursue versus pass over time and adjust scoring weight based on my decisions. Save coin weights to config and so they can persist across runs."
So this is actually a visualization of the deal flow loop we've created. What this version on our repository does, it, it first pulls the recent results of the deal flow we have it pointed to. We usually give it context and we say, "Hey, for our firm, here are the boundaries of the experiments you're allowed to run. What kind of offers you're allowed to make? What verticals we're targeting." So first, it pulls the recent results and it says, "Okay, what happened yesterday? What was the most recent experiment? Did it work better than baseline or worse than the baseline?" Then it rereads the context file to think about, "Okay, what are new experiments we're allowed to run?" And then it changes the filter, the listings filter, changes the broker outreach copy, relaunches the screening completely autonomously. It stores new experiments so we learn what west sourcing channels work and it constantly optimizes things. Look, if all those things hasn't convinced you to start using Cloud Code for your deal flow and acquisition strategies, I don't know what will, to be honest. Hopefully, we'll have another video coming out soon, uh, that will convince you of all the other stuff we're already doing. And again, last thing, if you don't want to build any of this yourself and you just want to get deals, get access to capital, you want us to make offers on your behalf and basically be your M&A team before you close on a deal and support you with AI before and after. We're going to put a link to book a call below. Either way, hopefully you enjoyed it. Subscribe if you did, and I'll see you in the next one. Let me know in the comments below what kind of things you want me to go through as far as AI and M&A. Hopefully you enjoyed it, and I'll see you soon. Take care.