No-code AI Workflows Every Business Leader Should Know | Pietro Montaldo

A recorded interview (YouTube) between James Hotson and Pietro Montaldo about no-code AI workflows for non-technical business users. Pietro—co-founder of Enforce—describes how he uses no-code agents and workflow platforms (e.g., Relay app) to automate content creation and sales tasks. Key examples include a Reddit agent that (1) identifies top subreddits, (2) scrapes rising posts from the last week, ranks them, and (3) drafts ~10 contextual comments for outreach; Pietro splits his work ~60% content and ~40% sales, relies on OpenAI models as a base while mixing other providers (Claude, Anthropic) by use case, and predicts rapid progress in image/video generative tools.
Questions this video answers
October 14, 2025
Approximately 60% content creation and 40% sales
Around 20% (as cited during the talk)
Searches rising posts from the last week and drafts about 10 contextual comments per run
Relay app
Chapters
AI sounds Lopez when I didn't do a good enough prompt, right, which happens 60% of something I would like every time I'm developing something new. Right.
Hey, I'm James Hodson and I'm here with Pietro Montaldo, an entrepreneur and co founder of Enforce, which is an orchestration AI platform that he designed to make artificial intelligence more accessible for non technical users. He's based in Barcelona and he really focuses on sharing practical AI tools so that anyone can take advantage to create real world solutions. And the key there is not needing any technical background with all the stuff that he does. So he brings a unique blend of business insights and product thinking and he's really helped me personally understand the fast evolving AI landscape. He's always sharing great nuggets of knowledge on LinkedIn. I found a recent post by him which was an AI guide to Reddit. Really useful and it's driving an insane amount of SEO for us. So welcome Pedro. Thanks so much for taking the time to talk today James.
I'm very glad to be here and I'm really excited to share my insight. I really like to talk about what we do, so I was very glad you invited me on this series.
Amazing. Give us a quick backstory about yourself.
Real quick story. I think it's mostly interesting to talk about what we're doing but uh, my co founder and I have very different background. I come from a business background, have started my career in VC and have been a go to market operator in the last five years. I led the sales team for a startup and I was mostly operational and on the revenue side. Then while doing that fell in love with, with AI. I was trying every tool when I started. No real tool was, was actually working. I tried AISTR 11x when it was just launching and I started geeking about these tools. In the meantime I met my current co founder which is a software engineer with 25 years of experience and we launched what started as an agency to develop custom solutions and evolved into an agency plus a product that really helps make AI accessible for small and medium sized businesses.
What was that light bulb moment? You're like oh, AI is going to be big. This is how companies can use it. When did you first realize that was that transformation was possible?
I think it's, it was when I could actually build something useful for myself without having to rely on on product people. So I think, I think the tipping point when I started using workflow, no code workflow tools like Relay app, lindiorrelevance, which was approximately a year ago now. When they started being really accessible and I started building really cool stuff stuff to handle my email, stuff to handle my content. And I was like, oh wow. Actually I can do stuff which is production ready where production is my own environment, but really amplify my possibilities. And then of course I mean the chat GPT moment and chatting with the. Having it live on my phone at all times for sure.
I was like, okay, okay.
This is, this is the first major technological shift I can actually remember in my conscient life.
Yep, I am with you on that point. I can't. I was talking with VP of engineering this week and he was like it's pretty amazing to see how quickly things have shifted and how people who have no technical expertise are able to deploy technical solutions. Now before we jump into like some of the ways a business could use it, I'm curious, what are some of the ways you're personally using AI for yourself in your day to day? What are some of the things that you're doing that maybe not everybody's doing?
So my, my, my personal work has divided, let's say 60% content creation and 40% sales. My partner handles all of the complex implementation that requires coding. So I use AI and both of the section of my work. Number one content creation, my entire processes orchestrated by me. But the hard work is done by AI ideation to post refinement, to post drafting, to finding hooks for social media. I have basically one GPT for everything which is exclusively specialized on the specific activity up to the posting and repurposing. I was, I was, I was telling you before. I'm trying now to expand my audience from LinkedIn to Twitter and YouTube and AI really helps me repurpose the content with what works on another platform or use the idea and the content I've already produced to generate YouTube videos. Right. So this is a part which I think people are mostly comfortable. Everything text based people do and then everything else which is a bit more counterintuitive is not counterintuitive but let's say less widespread is everything related to the sales part. So preparing for sales call and funding in interesting information to be able to relate with people. Understanding exactly how to sell with people based on their psychological profile. Scraping all of the information from, from LinkedIn and preparing me a brief that I can take five minutes before the call and I have like a cool opener. I know the study at the same university went on holiday there and I have like some cool information to break the ice. I think that's, that's a really Cool thing we do and everything which is post say so preparing a follow up summarizing what we discussed and the next point that's also entirely automated and recorded in the CRM. So I think it is slightly more complex but non technical user can actually do it and I think probably less people are doing it, but it actually brings more, let's say just a bit. It's one step farther than the content which is just chatting with a chat interface because it also interacts with CRM and other systems. So that might be scary, but it actually brings the most advantage.
Yeah, I get it. And meeting prep used to take me an hour, maybe an hour and a half sometimes. Now that time window is shrunk and I've just like been tuning my prompts, asking it slightly different things so I get more of the information I'm really wanting every time I'm doing that meeting. So maybe now it takes me 10 minutes versus an hour and a half previously. So it's, it's a tremendous time save if you're doing four or five meetings a day with different people.
Yeah, absolutely.
I believe you have some stuff. Would you be opposed to walking us through some specific use cases that you've built?
Yeah, yeah. I mean since you mentioned it before, I thought about sharing the Reddit agent which was very popular on LinkedIn and people really liked.
Yeah.
The concept behind this agent is that Reddit makes up or made up because I've been hearing they have been changing something how they index Reddit content on Google. But up to last week, let's say Reddit made up to something around 20% of LLM citations. Right. So if you want your product to appear on LLM responses, you'd be a fool not to go and promote it on Reddit. Right. A lot of people don't have so much time to spend on Reddit finding the right subreddit to. To. To post on. To find the right post in the right subreddit to have opportunity to comment something meaningful which doesn't sound too salesy about their own product. Right. So the idea I built two separate agents. The first agent was to identify some subreddit for your startup. Right. So you would feed the agent information like your, your website for example, something super basic and it would help you identify key subreddits where you where there might be topic which are relevant for your startup. Right. It would identify the top 10 subreddits for your startup and you could then select which are the most important one. And I'm going to show the second Agent, which is the one you see on the screen because it's slightly more insightful instead of like once you have identified the subreddit, you have to go in and check what are the posts which are actually relevant for your company at least where you can add some, some sort of value, right? So basically here what this agent does and Speed Relay app, which is probably one of my favorite platform to build no code agents. So here I fed them the two subreddits, which is mostly interesting for my startup, right? So you see product marketing and product entries. The agent knows where I want to look and then I fed again the information about my company. In this case I'm using a third party component just to demonstrate which is Relay app, which is what we are going to build this flow, right? So if I had the company URL, the company blog and the link and the URL. So the agent has enough information about what we are and what we are doing, then what we are doing is basically create a comprehensive description based on all of the information we fed so that there is a profile about my own company and the agent knows the, the, the data points to check for on Reddit to find relevant posts. Now it's where we get to the Reddit part, this part search for key Reddit posts. So basically for every subreddit which identifies. So for product marketing and product hunters it is agent iterates through and looks for the rising post which is an important key information, right? I'm not looking for posts which are already big, I'm looking for posts. We're getting a lot of engagements, right, of the last week, right? So I'm looking at the last week and posts which are rising in the last week, right? So they are still at the beginning of the life cycle, so more people will see it and they're going up, they're rising, right? So it's a perfect type of post where to leave a comment. And then once we have found this post, we have an agent that looks specifically for all of the posts which we extracted for Reddit and looks at the one where we could add the most value with the comment. Not only that, but it also drafts a comment so that you can review. Now the draft is never perfect, right? It will always sound a bit like AI. This is the keynote ones. You have everything up to that. But say you want to run this agent once a week, say on Friday at the beginning or Monday at the beginning of the week, you check everything that was rising up to the last seven days, you take those posts and in 30 minutes, you leave 10 comments. You know, that makes a difference. The comments are insightful. Let me show you actually how it looks. As I said, this is. This is a trial run, which we did last week or two weeks ago, and this is how the results would look like. I don't think you're seeing it. Okay, so this is a Google sheet results. And this is the first. It's ranked, right? So it tells me, okay, this is the best post you can comment with. So if I have five minutes, I can focus on the top five, right? So it ranks everything. It tells me the post title, the post URL. So I can go check it, why it was selected, why it's an interesting opportunity to comment. In this case, say it was a post about sales enablement, which is one of the core on which Rayleigh develop his agents. And it provides an answer, right? You see, the answer is contextualized with all of the information the agent has about me. And it can really, like, it's structured also in a way, like, the agent knows how I want it to be structured so that it's SEO optimized or IEO optimized. And this is already pretty. This is not touch, right? And it's a good comment, you know, okay, you can cast the double dashes from ChatGPT. You can, you can reformat it a bit, but it's. It's 95% there. It needs a final touch, maybe at the beginning, at the end, and you can post. And you have 10 of them, right? And you can have more. I just told him to create 10. And this can save you an incredible amount of time. And you can open up a Reddit channel, a Reddit growth channel for your company with what it would have taken you, I don't know, a year and whatever, three, four hours a week. It takes you 20 minutes per week, and you get better results.
I love this. So, you know, the traditional way, if you weren't having this automated, you're spending three, four hours a week doing this manually. And like you're a human, you're getting distracted, you might not do it the same. You might. One week, you might do it really well. One week, you might not do it so well. And then if you have employees doing it, like, I like that it kind of does the same thing. How are you?
I'm.
I'm also struggling with the. I call it AI slop, right? Like, it's really obvious. The AI generated text and copy has a very distinct personality to it most of the time. And so I spend a Lot of time now editing those things. And I've even put some commands in my prompt that say never use EM dashes. You know that because that was a big thing. And then start to format it like in my tone and voice. Have you found anything good for getting that content to be on brand for patrol posts like me essentially? Or you know, put this messaging for LinkedIn style versus Twitter style versus, you know, your other channels.
Yeah, I think like what I see from myself, I'm not saying you as James, you do it. But when, when AI sounds lope is when I didn't do a good enough prompt. Right. Which happens 60% of something I would like every time I'm developing something new. Right. If you put the right amount of time in reviewing the prompt before using it and I also always keep that part, it will sound like you. You give them two, three examples of stuff you've built. Tell them building like this, at times it will have a little bit of a slope, but it's an exception. Right. If you do a good enough job in prompting, the AI can sound like you or very similar to you or less sloppy in general. And there are so many LinkedIn tools today. A magic post is the major one. And from there and just like Cleo and. And then one pops every week. Right. They do a decent job. Right. Nothing will bring you impression and visualization out of the blue, but you can just pick your own poison. Personally, I think you can achieve almost the same by yourself by doing a nice GPT agent or cloth. But I also spend whatever four or five hours a day in the last six months. So I figure not, not everybody has that time to allocate because this is not just priority for them and going to one of the provider might just be a better solution.
Yep. The more and more I talked like you got to do the reps yourself, you got to figure out, you know, what are the tedious tasks that you can start to automate that AI can get plugged into and save you time. It's like augment, so you can do 10x the amount of work, not rip and replace that. This one magical AI is going to do all these things.
That's our mentality.
I don't know if you saw this, but yesterday it was announced OpenAI had their agent builder and that's kind of been the talk of the town, or at least the talk of the net. What, what do you think about that? Do you think tools like N8N and Zapier start to struggle or do you think like, you know, we can. It's like a rising tide raises all boats.
I think it, I think the second one right. Partly because I want to believe it's a second one and, and partly because I haven't tried it. Right. So I like I'm going to try today, tomorrow I just didn't find the time yet. I will try it and be able to give you a better impression. But in general there are platform who have been working on education and customer experience and UI and UX of what you're building and are really, really good at it. Right. OpenAI has released a bunch of things that yeah, okay, it's fine. Nobody used. They have an incredible advantage that if somebody has only chatgpt, OpenAI and doesn't want to download anything else, sure, you know, they're already paying, they're probably going to go there, you know but I guess when they get to they need to make it more professional and more like widespread in the company. I think the companies that are specialized on that will, will prevail and in general it brings more attention. More people start building more people want more complex stuff which they can do. ChatGPT, they go outside and they go to Zapier and I turn relay of the of the sword. But again I haven't had the chance to think about it. Maybe it will sn a bunch of, a bunch of code. Agent P this is my first impression, my first impression good for everyone just brings more people more interested and then also I mean for me doesn't really make a difference because we can play with any tour, you know, like next week I'm sure or even later this week I'll make a post on using Agent Kit and showing that around and if it was better to bring, you know, more clients for us, I'll keep doing it.
Right. Yeah, I mean it makes sense. I think you're spot on and I really agree with the niche tools that are purpose built, you know, will still be around because there's so, there's so much thought and so much nuance in building those kinds of things, you know and we're starting to see the shift to smaller specific models as well. Are you using different models? You know, I'm paying for probably two or three different AI LLM type subscriptions right now. Are you using Gemini for posts and Claude for code and you know, chatgpt for daily stuff like what's it or do you just kind of commit to one model?
Yeah, so my base is on OpenAI. Right. I just started there and probably for the same reason she Told before I'm inclined to use a has ads it's always one in the top three. You know like they play a bit who's first but it's always one in the top three. So like that's, that's where I have most of my stuff pre built and I use constantly. It's true that Claude was better up to some time ago and to sounding more like you and creating like text content and probably it is a bit better but it's potato, potato. You know all of them are very similar unless you are doing something like for example for coding. I know a lot of, a lot of people love anthropic for coding and it just has no comparison with the others. Right. So there are some use cases in which it makes sense for the other more or less perplexity. It's great for a search but it's also true that when you go to third party automation platform like relaying the relevance of the sort of you can purchase directly through them. Right. And if you want visual you go to Freepik and you can have Gemini Visual cling and anything you want, right. So there's basically aggregator where you can have everything and you should repay a little premium but you can, you know, you can use the best thing for the best use case. So when I build agents which I share or I, I do for clients, we use anything which is, which is best and gets us to the best possible results. Right. We, we have no, we have, we have no shield on the.
Yeah, you gotta be. I feel like you gotta be flexible and agnostic these days. Are you, where are you seeing AI either like overhyped or just not working at all. Like what. Where are people failing with AI that you're seeing?
So I have, I haven't seen anything in outbound outreach that really works autonomously inbound. Maybe people already have interest. It doesn't matter if a human respond. They probably notice it's AI but it doesn't matter outbound, be it voice, sms, anything I yet have to find. And it's one of the largest things like there are like countless startup raised. Wow. Millions and millions. I haven't seen one that is worth the millions they have invested in research. Right. I was like 11x was one of the first one doing AI SDR. I think I can build something which is equally as good. You know. Now, now I'm not saying obviously like, but I don't know. It's probably the largest because every company wants more leads and it's the first one thing they're happy to invest for I have to see one especially voice. It's not there yet, but even just email instantly doing a really good job. But most of it humans still have to do it. You build a copy just like the human is still in the driving seat. And there's just like part of it automated, but it should be one of the first thing entirely automated because like everybody will pay big bucks for it.
And it's also extremely tedious and repetitive. So you think that it'd be prime for, you know, AI disruption, automation disruption. But I think there's also a reason why on the human side there's human sales reps that hit quota or above quota every single month, every single quarter and then there's reps that constantly underperform. It's like there is a nuance and a relationship building like you mentioned at the beginning of the call that has to happen in some sort of like seeing around corners for you to think about, oh, this is why I'm going to prospect this group or target this audience or this is going to be my message to this person. And it ends up being one to one a lot of the time. And yeah, I agree like we have, we tried AI outbound and got zero results. We went back to doing it pretty much manually and are having much better success and actually getting more meetings. So it's, it's interesting to me. One of the things that I've been thinking about quite a bit is like how much electricity all these models are consuming with everybody prompting all the time. And I saw Nick Huber post about his electric bill in Atlanta going up 60% because they're building all these data centers and all this compute and we're kind of just like I'm guilty of it at times, like mindlessly just prompting and talking with GPT or GROK or whatever. Do you have any thoughts about, you know, how much energy and electricity this is really costing us? Is it, is it a net benefit with all this AI stuff? With all the electricity cost? It's, it's kind of a crazy environment right now.
Honestly, I don't, I don't really have an opinion. I could repeat like a para thought I read on Lincoln because I spent a lot of time there. I mean I, I think that is a big market in reducing the cost, the electricity cost of LLMs. Right? And I know for a fact they're investing big buckets into finding ways to reduce it. Having said this, honestly, I just don't have a good enough opinion for me to claim, to claim something on the.
Topic, I know I, I noticed myself starting to think these days, like, oh, is this a perplexity query or is this a quick Google search? You know, is like, what, what's the greener way to do this thing?
Yeah.
So if we were to fast forward 6 months, 12 months, do you have any predictions of what AI will be doing in a year from now?
Every prediction I've done so far was wrong. Right. If I look six months ago, I was wrong. So I'm not sure I'm the right guy to ask for. I think it will be easier to build with only text. Right. And companies will go in that direction and we'll have like a first. It's already there, right. But it's getting better and better exponentially. Like you can build flow and micro app only with text and then refine them with the, refine them manually and keep a human in the, in the loop while doing all of this. And I think honestly the most impressive advancement and the fastest is everything which is video and image. You know, like every two months you have something like I wouldn't be surprised if you can have like a fully cloned version of yourself and full, fully a completely recognizable UGC accessible for $2020 a month. Right. Which is scary because I'm like, you know, I'm doing a lot of content, right. So there's always an opportunity anyway. Right. I always keep an eye out, try, try all of these things and say as, as soon as one thing is good enough I will just, I would just bring it in and adopt it myself. But visual and like video and images, the thing which I think moves the fast and it's the most impressive and changes things. Right. Because now I can do really stuff which, you know, the same thing that happened for coding a building micro app will happen for creating really incredible video. You just have to have the imagination and, and you type it farther than that. You know, I, I claim like I, I take pride in being a very practical person and I am not one of those brilliant minds that can look ahead and predict the future and where technology is going to go. And I know my place is that and I try to, I try to never look past the three months and especially in this, in this particular market.
Yeah, I'm with you. I think it's on my to do list to Go play with VO3 by Google. We have some marketing content we want to build, you know, a trailer or a teaser with. And then that was my Monday, Today's Tuesday. I'm reading all about the Sora 2 release. And I'm like, oh, maybe I'll do this with sora instead of VO3. Because it just seems like it's just constantly improving, improving. I think it's almost already there. Like the top video AI producers or prompters are producing content that you might not be able to realize is AI generated. And I think that's kind of crazy. And so where does that go? You know, everything with AI seems to lead back to human imagination and creativity is, is very valuable and human connection is extremely. And relationships are extremely valuable, which, which is kind of funny that it highlights that stuff. So I really appreciate you coming, coming on. How do people get in touch with you or find you if they're. They're looking to follow some of your LinkedIn posts or, you know, check out Enforce.
I am, I am a active every day on LinkedIn. So they can, they can go follow and connect with me on LinkedIn and send me a text. I'm. I'm always super available. I'm launching this week my YouTube channel as well. It's going to be a slow burn. I'm doing one video per week for a bunch of time, but it will be insightful tutorial with me. So it would be like 30 minutes to 45 minutes to drive through the building of an agent, which is mostly for people who actually want to build something meaningful. You work out every video with. With something built and working, which I think is very cool. And I would say these are the best ways. I would say leave my Twitter alone. It has 24 followers for now and I still haven't figured out how it works and I probably won't invest so much time in it because they just hate it. LinkedIn would probably be the best way.
LinkedIn and YouTube. I'll put the links for people where we post this and really appreciate you coming on. I think got some really good stuff.
Thanks, man. I really appreciate it.
Thanks, Petro.
More from James

The 7 AI Tools You NEED in 2025 from a Founder Who Uses Them Daily | Jake Hay

This Causes Most AI Agents to Fail and Its Completely Avoidable | Kevin Denman

He Sold a Solo-Built A Privacy App with AI Here’s What Actually Works | Alex Calish







