Pricing in the AI Era: From Inputs to Outcomes, with Paid CEO Manny Medina

Contents

Intro

Manny Medina: You know, I think it was Seth Godin who said pricing is part of your story. You know what I mean? And how you differentiate in your market is your story has to be different, and if your story is different and your pricing is the same, your story ends up being the same. You see what I mean? This is why, like, Sierra doesn’t have pricing on their main page. Because they’re sitting down with each customer, and they’re finding out what’s important to them. Is it time to resolution? Is it that the customer resolves a ticket and then buys something? Is it CSAT? Is it NPS? What is it? And then you build a box around and you say that that’s your outcome function. It’s just the same as an objective function in machine learning. Like, you create this function of where you want everything to go, and then you keep iterating towards it.

Pat Grady: Greetings. Today we’re joined by Manny Medina, founder of Paid. Paid helps agentic AI companies make money by dialing in their pricing and costs. Manny experienced the problem he’s now solving with his previous venture Outreach, where he first encountered the challenges of pricing and margin management—challenges that are becoming top of mind as agentic companies move from experimentation into production. Paid provides the infrastructure that helps AI companies transition from simple activity-based pricing to more sophisticated value-based approaches, allowing them to capture their fair share of the value they create for their customers. In this episode, we’ll explore Manny’s contrarian views on AI pricing models, his framework for the four pricing approaches that are working today, why he believes specialized AI agents targeting narrow problems are printing money and what it takes to build a successful business in the age of AI agents. We hope you enjoy.

Pat Grady: All right, we’re here in London for a very special episode of Training Data with Paid founder Manny Medina. Manny, welcome to the show.

Manny Medina: Thank you for having me. This is awesome.

Pat Grady: All right, we’re going to start off with a high-level question.

Manny Medina: Okay.

What’s working at AI app companies?

Pat Grady: In the world of AI apps, these are your customers, so I think you have a pretty good sense of what’s working and what’s not working.

Manny Medina: Yeah.

Pat Grady: What’s working?

Manny Medina: So I think that we are right now—and if you were to take the analogy of hedgehog versus fox, we’re in hedgehog land right now.

Pat Grady: I love where this analogy is going. [laughs]

Manny Medina: So, like, if you take up a very narrow problem, and then you hedgehog into it and you become the best at that one thing, that is printing money right now. Like, everyone that I’m seeing …

Pat Grady: What’s a good example?

Manny Medina: A good example? For instance, I love what Quandri is doing. I love what XBOW is doing in your portfolio. I love what HappyRobot is doing. There are very narrow cases of problems that have a lot of people in it that don’t have a clear solution for software. Like, there’s no software to solve that problem. It’s people heavy, and it’s interesting because they’re not replacing people right now, but they’re replacing BPOs. So whenever you see BPOs having a big role, that is food for the apex predator that we’re seeing right now in the world of AI, which is AI agents.

Pat Grady: And for the uninitiated, what is Quandri? What is Expo? What is HappyRobot?

Manny Medina: So Quandri does policy renewals.

Pat Grady: Okay.

Manny Medina: Another one that is adjacent to it is a company called Owl that does—they’re in the insurance space too, but they review claims. They review claims data. Again, super tiny. Like, you look at it and you pass because you think that the world is really small, but when you look at all the amounts of claims that people do and the ones that get left over, you know, they’re taking care of those. HappyRobot is calling truckers on behalf of brokers. So if you want to send beer from Milwaukee to Boston, and you call a broker to find you a trucker, Happy Robot spins up 2,000 agents. Agents call the truckers, which is, you know, a guy and a dog, and they negotiate with each other until they book the load, and then they call the trucker all the way to delivery.

Pat Grady: And of course, once that Milwaukee beer gets to Boston, the Bostonians are going to send it right back home.

Manny Medina: [laughs] Exactly. But somebody has to move the beer around. So this is a thing. And XBOW, of course, does pen testing. And again, they took out the pen-testing agencies, they rank them, they look at the best, and then they run it for you. So these are, in my mind, very narrow problems. So XBOW doesn’t come and say, “I’m going to be a full cybersecurity solution,” which eventually they will be, but that you’re saying, like, “I’m just going to continuously bang against the door of every single app and backend that you have, and I’m just going to figure out a way to penetrate it. And that’s going to happen, you know, with the highest quality possible.” And it’s going to happen continuously, which is unavailable right now.

Pat Grady: And is the converse also true? The things that are not working or things that are too broad in scope at the moment?

Manny Medina: I think there is a lot of traction in things that are broad in scope in terms of, like, companies coming out and they’re doing okay, they’re quite well. Like, Harvey is doing quite well. And they’re not so broad in scope. They’re narrow into the legal system. But I’m excited for what Crosby is going to do, for instance. You know, they’re tackling the problem in the similar space but with a different approach, just like, full out replacing the lawyer for commercial contracts.

But if you go too broad, like for instance, AI SDR is a little too broad, and that encompasses a lot of things and means a lot of different things for a lot of different people, there is going to be swirl. So I don’t like to characterize it into what’s working, what’s not working. I characterize into what’s working now and what’s not working yet. You see what I mean? Like, what is going to be refined and eventually come through? Like, for instance, the one that I’m waiting for with bated breath is AI assistance for EAs. Who’s going to be my AI EA? And I think there is Lindy and there is Fyxer out there, but they’re not quite there yet.

Pat Grady: Yeah.

Manny Medina: You see what I mean? For instance, I love Fyxer, but I have three time zones that I work through. My co-founder is in India sometimes and, like, most of our customers are in Pacific zone. I’m here. And, like, getting that to work out just right doesn’t quite work for me. But if I had, like, one time zone and I had, like, one line of business, like if I’m a real estate broker, they are perfect, like perfect fit for that particular thing. So, like, big problem, narrow application, nailed it.

Pat Grady: On the EA front, we have a founder to introduce you to. He’s in stealth, so I can’t say his name.

Manny Medina: Okay.

Pat Grady: It’s coming.

Manny Medina: Perfect. All right. Love to see it. Because this is one of those problems that I’m just going to wait. I’ve had an AI forever, and I’m going to bite the bullet and just kind of wait for AI to catch up. So I’m ready, guys. Send me business.

Which markets will transform sooner?

Pat Grady: Do you have a sense for which markets are likely to be transformed sooner rather than later, and which markets are going to be more resistant to AI transformation?

Manny Medina: So I heard this thesis. When I came up with the idea for Paid, it was right around when Strawberry dropped. And you guys wrote that paper on the shift to cognition.

Pat Grady: Act o1. Thinking fast and slow.

Manny Medina: I remember where I was. Like, it’s one of those things that, like, I read the paper and then I listened to somebody reading the paper, and I remember where I was. There’s so many episodes in your life you remember where you were when that happened. Like, I remember exactly where I was when that happened, and it just hit me. And I was listening to—I think it was Mamoon from Kleiner who had this hypothesis that I actually fully disagree with.

Pat Grady: Fully disagree.

Manny Medina: Yeah. So his hypothesis is that AI is going to start with the highest-paying jobs because that’s where the money is, right? So it makes logical sense. You go and target the most expensive job and you displace those developers, lawyers, accountants, doctors. I actually disagree with that hypothesis. I think AI is… the highest paid people will buy AI as a side thing and ditch it with the same regularity that they ditch other things in their lives. I think AI is going to stick the landing where it actually takes over our role fully that nobody else wants to do. So for instance, nobody wakes up and wants to be an insurance actuary or an insurance adjuster.

Pat Grady: Our partner, Roelof Botha would beg to differ.

Manny Medina: Yeah.

Pat Grady: Perhaps the only licensed actuary in the world of venture capital.

Manny Medina: [laughs] All right, so I’m sure there is an exception to that. But these are jobs that are hard to backfill.

Pat Grady: Yes.

Manny Medina: So when Quandri does these policy renewals people, they’re replacing people that are exiting the business and they’re not filling it back on, right? The same thing was, like, nobody wants to wake up and work for a BPO banging out the phone. They do that job for six months on the way to a different job. So the turnover on these jobs are really high. So what I’m observing from my end is that the companies that are doing really well are addressing pools of labor that are either disappearing because of retirement or they want to do something else.

Pat Grady: Yes.

Manny Medina: Or they’re run by BPOs.

Pat Grady: Yes.

Manny Medina: And in that segment I’m seeing a lot of stickiness, I’m seeing a lot of expansions, I’m seeing a lot of growth and really good economics.

Pat Grady: Yes.

Manny Medina: Like, I can charge whatever I want. You know, my upper bound of my price is the labor cost, my lower bound on my price is my margin, and people are going to town on that. So, like, in terms of people experimenting with outcome based, et cetera, that’s where I see the majority of my traction. In the broader application, the reason I disagree with the rich job replacement is because everyone is going to go there. You see what I mean? If OpenAI needs another source of revenue or a company needs another source of revenue, they’re going to go after lawyers and accountants. Why? Because it’s hard and they pay a lot and they buy everything. So I feel like that market is going to be hot for now, and then it’s going to get super competitive and then you’re going to have a lot of people in it.

Pat Grady: See, I wonder if you can have your cake and eat it too. I wonder if both you and Mamoon are correct in the sense that for the higher-paying, more creative jobs, I think the copilot approach—and I don’t mean Microsoft Copilot—I mean the approach of AI that is giving people superpowers, seems to really be working. You know, Harvey in legal or Open Evidence and medicine. Whereas for the jobs that are lower paying and a little bit less creative, the full autopilot approach seems to be working where you’re fully replacing the work with an autonomous agent that can do it better, faster, cheaper. And so I wonder if it’s both. And then as far as moats for the copilots, I think you’re right that because those are big categories where there seems to be a lot of money, there’s going to be a lot of competition. I think what we’re also seeing is collaborative workflows, which are kind of the eighth wonder of the world as it comes to software.

Manny Medina: Right.

Pat Grady: You know, collaborative workflows still work, right?

Manny Medina: [laughs]

Pat Grady: You tend to get pretty deeply embedded into your customer, and then once you’re in there, you’re providing a ton of value and they don’t really want to switch.

You can vibe code a ServiceNow

Manny Medina: Right. No. So it’s true, but I think the world has changed a little bit in that I don’t know that you’re going to see Asanas anymore or ServiceNows, or any of those. They work in the collaborative workflow when it didn’t exist, and they were revolutionary. I think now anyone can spin up—you can vibe code a ServiceNow.

Pat Grady: This is a good topic. So vibe code, are you a believer or not? Where are you on vibe coding?

Manny Medina: I think we’re—dude, we spun up Paid in, like, a month and a half.

Pat Grady: You are a vibe coder.

Manny Medina: And half of it was vibe coding. I mean, we had to throw it away. But this is the beauty of vibe coding: You can throw it away and re-spin it. You know what I mean? Like, there is a lot of people getting hung up on the whole debugging thing. You don’t debug vibe code, you throw it away and you start, and you put it in production and when it breaks, you start. It’s wonderful. You can always start.

Pat Grady: This is very comforting to your customers.

Manny Medina: Well, I don’t have a lot yet, so I don’t have a big problem. But eventually, it will be a big problem. To start is a very easy thing. So back to the workflow question. I think that you’re right. So if you become a definitive workflow for X, Y or Z, like, it’s definitely a sticky point. I just think that the speed of competition is really high right now.

Pat Grady: Yes.

Manny Medina: And copilots have—the explainability of the value of a copilot is really hard to land. Like, how do I differentiate one copilot versus the next versus the next versus the next? And they all can come and say, “I’m the same as X but better or cheaper.” And that’s a recipe for swirl. So unless they specialize in verticals and say, like, if you are—you know, if Harvey were to say for, I don’t know, patent law, “I am the best and I got 78 percent of the market,” boom, right? But now you’re moving into the very narrow application with very narrow set in a very rich market. It’s rich because you own the whole of it, not because the whole market is rich. You see what I’m saying? But look, we’re all hypothesizing here, like, I make money either way, so I’m happy for all of it to work out.

What’s working with pricing and packaging?

Lauren Reeder: So you’re trying to jump in to help with this on pricing and packaging. What are you seeing work today?

Manny Medina: So I’m seeing four things sticking the landing with aplomb, with gusto. One is, you know, clearly charging by activity. You know, that’s an easy one, has a credit consumption-type model, and you can show the activity that was done fairly easily. The other one that I’m seeing more is changing by workflow. When you string a number of activities and you say, “This workflow cost this much.” Like a document review, right? Because then you can separate documents that are small from documents that are long and complicated because they have different consumption patterns.

Pat Grady: And it feels like you’re getting closer to value-based pricing as opposed to cost-based pricing.

Manny Medina: Exactly, exactly. So moving to a workflow allows you to move out of the treadmill of charging for pure work to charging for work that is worth something to somebody. And then eventually you will get to some kind of outcome. And what I’ve been recommending to my customers—not out there yet, but I’m going to give it a push—is instead of charging per outcome, get an outcome bonus, meaning if an outcome happens of a particular quality that is measurable, charge for it. That way it opens the door to a conversation of value alignment. And the moment you open the door to a conversation of value alignment, then you start getting into more bespoke contracts with each of your customers which are super hard to rip out.

Pat Grady: And historically, that’s been a tough pricing model to pull off. Do you think AI changes that?

Manny Medina: I think AI changes it completely. Completely changes that, because in the past we wanted to put everyone in these little boxes called SKUs, and then we wanted to count SKUs, and each SKU will have a discretionary amount of discounting and whatnot. That was a world of rows and columns, and you don’t need that anymore. If you go to the largest companies, again like a ServiceNow or a Salesforce, all their large contracts are bespoke, you know? And you send off Paul Smith or the CRO to go and sit down with their counterparty, he comes back with a deal. And the deal has all sorts of complications in it that doesn’t belong into the CPQ world, you know what I mean?

And I don’t know why in the agentic world you wouldn’t do that all the time with the customers that you want to go big with. And you can always put a chat interface to say, “Interpret this contract for me and give me the annualized value.” You can inquire all the body of contracts that you have done and get a sense for, like, your unit economics and your growth and what this looks like. So I think that custom contracts is here.

Pat Grady: Okay, you had four things. There’s activity based, there was workflow based, was outcome a third?

Manny Medina: Outcome is a third

Manny Medina: And what was the fourth?

Manny Medina: The fourth is pay by agent.

Pat Grady: Pay by agent. How does that work?

Manny Medina: So I’ve been working with a lot of the AI SDR companies to get into, to introduce this as a concept, because a lot of what they’re doing is replacing, say, 80 percent of what normal SDR would do. And an SDR fully loaded will cost you somewhere between $70 and $90,000 a year. So you can pay—instead of saying a platform fee, say, like, “I’m going to deploy X many agents, the agents are going to do this amount of work that is equivalent of a $90,000 a year SER. I’m going to charge you $20,000 per agent. The agent is going to deliver this much work, and you can pay me a bonus for meeting both.”

Pat Grady: Yeah, I guess how do you define the job of an SDR? Do you just say, “Hey, your human SDRs have a certain quota. This agent’s going to hit the same quota.”

Manny Medina: Yeah. When you hire an SDR, the first thing you get is activity.

Pat Grady: Yeah.

Manny Medina: So you get X many calls, X many meetings. You have a book of accounts, you have contacts within that account, and each of those contacts gets an activity, right? So that’s where you get the 100 calls a day type of SDRs. You can do the same thing with an AI SDR. You can say you’re going to draw a boundary around the activity that this agent is going to do, and then you’re going to charge for that agent. Because the output is the same, right? And then your job is to say, “Look, instead of paying $90,000, pay $20,000 for the agent, and then pay me a bonus for meeting book; pay me a bonus for the opportunity to close one.” Which is very similar to the equivalent of hiring, which is the hiring of a human in the flesh SDR. And you get to dip into the headcount pool of budget, as opposed to the tools pool of budget. So you’re not in the RevOps purview in terms of where my budget is coming from, you’re on the HR, which is far larger. So that’s why I’m sort of trying to steer people away from charging like it was a tool, because then you’re constrained, right? Then the CRO has a budget for so many things, and you’ll get a sliver of that as opposed to full headcount replacement.

Pitfalls of charging as a tool

Lauren Reeder: What are the pitfalls that you’re seeing with companies that are ending up in the charging-as-a-tool bucket rather than charging for the work?

Manny Medina: That then you get pigeonholed into seats.

Lauren Reeder: I guess, like, what are people doing that makes them get pigeonholed?

Manny Medina: A lot of what I’m seeing selling is the vast majority of AI agent companies are doing PoCs right now. I mean, they look like contracts and they look like their money, blah, blah, blah, blah, blah. But in reality, every company on the planet’s got an AI mandate. So somebody in that company went and, you know, pursued some software, and then they bought it, and they bought it as software and as a trial.

Pat Grady: Vibe revenue.

Manny Medina: Yeah, a hundred percent. Okay, so now you have the vibe revenue curve. And now soon we’re going to come into renewal land, and that’s gonna separate the wheat from the chaff. And at that point you’re going to figure out who’s got real stickiness and who doesn’t. And that’s where the real monetization scheme is going to come in. Because they’re going to figure out, like, who made money, who didn’t make money, what kind of value they deliver or not.

Lauren Reeder: They’ll see the outcomes.

Manny Medina: And then we’ll see the—right. Exactly. A hundred percent. And it’s funny because, you know, a lot of my early customers, I told them how I thought about the world and it’s like, “Yeah, that will never happen.” And now they’re calling me back and being like, “Hey, I just got my first contract that only pays me per outcome, and I’m calling you because I don’t know who else to call.” So I feel like this is going to be pulled by the buyer. You see what I mean? As a way to mitigate risk. Because AI is risky in a number of dimensions, and outcome actually lowers your risk.

The pricing maturity curve

Pat Grady: In the four things you mentioned—activity, workflow, outcome, agent—is that a maturity curve of sorts? Like, is the idea that you want to get to selling agents or maybe you want to get to selling outcomes?

Manny Medina: It’s a fascinating question. So the answer is I think there is a maturity element into it, but it’s a little bit of making your own adventure element to it in that there’s a bottom line. So everybody has to get out of selling by activity. If you stay there, somebody will come along and say, “I’ll do the same thing for cheaper.” And then you are in a nightmare scenario in which there is tons of others who look just like you. messaging is just like you. And the only way to separate the wheat from the chaff is to try.

Pat Grady: Yeah.

Manny Medina: And now you’re, like, churning from one to the next. So yes, there is a maturity level in that. If you don’t move out of that bottom one, which is easy to sell, you’ll get competed out. Once you get into workflows, then you’re into value-based pricing and then you’re defining what the workflow is and why is that important. And then you are in a deeper conversation, and that forces a much better alignment. So I guess the answer to a quick question is yes, it requires some level of maturity to have that conversation with your customer.  Your customer will always default to the easiest way to buy, which is either some kind of fixed price or a consumption price for the first year to see if it works. But if it does work, it is up to the AI agent builder and creator to go back to the same customer and say, “Let’s align on things that are important to you and charge for it.”

Pat Grady: Do you have any sense of which markets are likely to stay with value-based pricing, and which markets are likely to collapse into cost-based pricing? Have you seen sort of early indications from the companies that you’re talking with or working with, you know, of markets that are trending in one direction or the other?

Manny Medina: So what I’m seeing right now is that for those who are targeting BPO budgets …

Pat Grady: Yes.

Manny Medina: … to win that business, they go at the BPO pricing, lower it, and they say, “I’m going to do the same as a BPO, cheaper, 24/7, and I’m going to accumulate all the data that the BPO used to do.”

Pat Grady: Yep.

Manny Medina: But again, I’m seeing that as an intermediate step. Like, that is not the full step, that’s just a way to get into the market and win it. Then I wonder, the BPO is not going to just sit there and, like, just take it. You know what I mean? Like, these are relatively large companies, and they can just as much buy somebody’s technology and deploy it. So I wonder what will happen once the BPO turns around and deploys their own agents, replaces their own people, uses the data that they have internally to train or to make it better, and then go to town and defend themselves.

I was actually talking to BCG about this, and they wouldn’t tell me that they’re doing a lot of engagement with BPOs, but I sort of sensed it. And I don’t know that they’re just going to sit on their hands and see their business go away. Like, that will never happen, right? So I wonder how does the game sort of like sort of levels, if you would. Like, where is the balance of trade? I don’t know the answer to that. But I think that to get into a market, you see a lot of initial, you know, let’s price it the easiest way for somebody to consume. But to advance into the market, you immediately switch to something else. Like, in the AI SDR, you see that everywhere. Like, everyone came in and charging by some kind of token or credit or activity. But then it became a bloodbath and there’s like 50 of these guys, and they all kind of sound the same. So unless you start making some assertions of your quality and putting your pricing to back that assertion, meaning I’m going to get you five qualified meetings or I’m going to charge you per agent that replaces a human being, then you’re going to get, you know, competed away.

Pat Grady: Yeah.

Lauren Reeder: And as you move into the later maturities of pricing models, how do you see people measure and actually implement these things?

Manny Medina: I think that’s kind of the beauty in that to each customer, the definition of success is going to be a little bit different, and a little bit of like, I feel like I was like John the Baptist walking around like eating locusts and honey and, like, looking like a crazy man telling people that they should do bespoke contracts. And everybody was looking at me like I’m crazy. Now it’s coming around to say, like, you know, the ability for you to understand your customer’s business and price and contract around it and have the engine behind it to support it is actually a competitive advantage in a world in which you all have the same tools.

But it wasn’t obvious. It wasn’t obvious when I started. I remember we talked in September last year, and it wasn’t obvious back then. It’s getting a little bit more obvious now as we’re turning on the lights in different parts of the market where people are saying, “Yeah, we need to do something that is very unique to me, so that I can protect that asset or that contract.” Second of all, you know, I think it was Seth Godin who said pricing is part of your story. You know what I mean? And how you differentiate in your market is your story has to be different, and if your story is different and your pricing is the same, your story ends up being the same. You see what I mean? So aligning with—and this is why, like, Sierra doesn’t have pricing on their web, on their main page, because they’re sitting down with each customer and they’re finding out what’s important to them. Is it time to resolution? Is it that the customer resolves a ticket and then buys something? Is it CSAT? Is it NPS? What is it? And then you build a box around and you say that that’s your outcome function. It’s just the same as an objective function in machine learning. Like, you create this function of where you want everything to go, and then you keep iterating towards it.

The cost side of the equation

Pat Grady: Yeah, well put. All right, talk a bit about the cost side of the equation and how margins play into this.

Manny Medina: So it’s really interesting because this is another one that I’m contrary. Everybody’s telling me that the cost of tokens is going to go down and this is going to become a commodity and whatever whatever. I think in a world of reasoning, where is inference time compute worth more than training time compute? I just don’t see how that—at least in the immediate future, I just don’t see how the token price goes down. Because you’re going to require, like, deeper level thinking. So what the problem with agents right now is that you sort of like workflow them, right? So you go to LangGraph, and you put your boxes and what each does and you string them together. But the fundamental problem with that is if you have a hallucination at the very beginning of that chain, you’re screwed. You have all sorts of, like, bad activity happening all through.

Pat Grady: Yeah.

Manny Medina: And there’s no amount of eval that is going to rescue you out of that. So the better way to do it is to have the model do most of the work. You see what I mean? As opposed to have these little boxes of work, have the model do most of the work, have a good eval framework to make sure that it’s doing the thing that it’s supposed to be doing, and then roll it out. In that world, where you have less error rate, I don’t see the cost per token going down. I see that if nothing else is going to go up as the models get more advanced and smarter.

Pat Grady: Or it could be the horse race between volume and price, and price for a fixed workload may go down over time, but workloads increase because you’re throwing more compute at things at inference time or because you’re just doing more sophisticated jobs.

Manny Medina: Yeah, I think over the long arc of history, I think you’re absolutely right. I think in the very short term, as we’re trying to figure this out, I just don’t know is sort of the short of it.

Pat Grady: Yeah.

Manny Medina: And the second thing that I’m seeing actually, which is my big a-ha moment, is that agents are using one modality, and the moment you nail the modality, you want to use all these other modalities, right? So you do text and then you want to do a phone call or to take an inbound call, and to do an outbound call or do an avatar or whatever that is, and then you’re incurring costs that are not LLM but are third party APIs.

So the full cost of the full service, you got your cloud cost, you got your LLM cost, and then you have all this other stuff that you have to buy to make it sing in the different modalities in which it’s used, or even buying data or buying whatever. And that is what’s driving up the cost. So if you look at, say, an avatar company or a phoning agent, their cost is not quite the LLM. Their cost is how long are you going to stay on the phone, and the number of dials that you do, or how long does the avatar run, et cetera. Those are compounding costs that you have to look out for. The problem that we have right now in cost mode is that because everything, like all the activity of the agent goes down through this eval framework that acts as a proxy before the token goes to the LLM, you don’t know who’s incurring what cost. So you don’t know what customer is profitable to you, what customer is negatively impacting your margins. You don’t know what agent is doing a good job, what agent isn’t. You see it as a whole. You see what I mean? And this is why the margin problem is such a bear to solve.

Pat Grady: Well, and generally speaking, margins are a reflection of the amount of value that you’re providing to your customer. Do you see that in AI apps right now, or is there a mismatch between the amount of value that companies are providing and the margins that they’re able to get?

Manny Medina: It’s a mismatch because people don’t know how to price and people don’t know their cost. So this is why the problem compounds a little bit in that agent companies are relatively new, so they come to the market just trying to, like, get business, right? So I’ll price in whatever way you want to buy, and only later I find out whether I win or lost. You see what I mean? So I think that this is just the beginning of the game, that as they understand the unit economics, they’re going to price better. As they understand the value delivery, they will price better, but they just don’t see it yet. So for instance, we build this thing in Paid that we look at an activity or a workflow, and then we make this call to the service that tells us what is the human equivalent of the same work in what country, right? So if the work is kind of hard, it looks up the type of work, how much would that person make per hour, and how long does the work take and returns a value.

So we give our customers this little bit of pricing guidance that says the human equivalent of the same amount of work is this. You can price up to that point. You see what I mean? Or at least use that to go and say, “I need an increase in what I’m getting paid because this is actually pretty expensive.” But without that, it’s kind of tricky. And it’s not like you have a Simon-Kucher pricing specialist alongside you, guiding you all along. They come in once a year and then they peace out.

Pat Grady: [laughs]

Manny Medina: You’re left holding the bag. Whereas you actually need guidance on this stuff every time you talk to your customer to get sure that you’re getting your price’s worth. The problem we’re seeing is that the value is accruing to the customer, not to the agent business. They’re capturing all the value of all the savings, and that needs to change.

Pat Grady: Yeah.

The mission behind Paid

Lauren Reeder: So you started a company to help this. Tell us a little about why and what your mission behind it is.

Manny Medina: When we were rolling out agents at Outreach and I wanted to understand the business fundamentals of the agents—my margins, the value that I’m adding, and how am I adding value—the underpinning software supporting me was not helpful, was not built for a world in which pricing needs to evolve, where the agents are delivering more than just bits and bytes, they’re delivering these full outcomes. And that just stuck in my head as I moved to London and I stepped aside from being CEO. And I sort of like noodled on that. And the problem kept on bugging me. So I’m like, “Is this a problem that was just me, or who else had this problem?” So I spent a couple of months just calling friends and other founders. And I always have this trick. I know a lot about sales and they know a lot about agents. So we will do a cross education. Ask me—like an AMA, like, you ask me anything about, like, building out sales teams. And I asked him about, like, what it’s like to run an agent business. And I found that, like, running the business itself was all spreadsheets. And that’s number one.

Number two, some of the problems were actually difficult, but not intractable difficult. There was just a lot of work. You see what I mean? So I love when there is a problem that is big, people are hacking around it, and the problem is not simple as a couple lines of code kind of solves it. So that sort of was the inspiration behind it, solving the problem. And number three, I wanted to build a company that I get to work with, like, the new founders because they were building a company themselves. And that kind of energy just drives me.

So the most fun thing to do when I was running Outreach was customer conversations. But the customer conversations at the founder level are so raw, are so full of, like, wonder and mystery and energy. And I can just see my energy tank just going up and up and up in every single one of those conversations. I’m like, I want to do this for a living. This is a great place to exist. And the market is going to be big, and there’s a small enough market that I can just cold call everybody and, like, most people will return my call because it’s founder-to-founder stuff, you know what I mean? I see your problem. Maybe I can help you. Maybe you can help me. And boom, you’re in a conversation. So it didn’t almost feel like selling, it felt like I was solutioning. And at the end of the day, I told them where I’m building this thing, and they were like, “Yeah, let me try it.” And I had a hundred percent hit rate on people who said, “Let me try it.” I’m like, “Oh, shit. Let’s go!” The worst that could happen, somebody buys something. You know what I mean?

What is Paid?

Pat Grady: We haven’t actually said explicitly what is Paid, so maybe just answer that question. What is Paid?

Manny Medina: Yeah, thank you. So Paid is the business engine for AI companies. So what we do at Paid is that we build the entirety of your billing, invoicing, monetization, pricing. Even at this point, we’re doing collections, revenue recognition. So the entirety of your vendor management and margin management. So the entirety of the back office you need to get your business up and running, to understand the unit economics and run your business, that’s what we’re building at Paid.

And our first foray into the market was our monetization engine and our margin management engine, which is the biggest problem that we’re seeing in the market. But we’re building the adjacent, so the problem compounds, right? Like, the moment you solve monetization and invoicing, then you need to solve collections. The moment you solve margin, then you need to solve vendor management, and so on and so forth. So we’re just building the unified layer that allows you to run your business in one place.

Pat Grady: How much of what you learned at Outreach transfers directly to building Paid, and how much is new?

Manny Medina: The human aspect is the same. Building teams is the same. Inspiring is the same. Selling ahead of capability is the same.

Lauren Reeder: [laughs]

Manny Medina: What’s new is how savvy the world has become in terms of company building. I think you made this joke in our podcast that you look at an AI agent company, you take the hood off, and it’s just building a company.

Lauren Reeder: Yes.

Pat Grady: [laughs]

Manny Medina: And the building company part has gotten so much smarter. Everyone that I talk to is way more advanced than I was at the same stage in my building of Outreach, where people have seen what happened in SaaS, the whole arc, and they’re building completely differently.

Pat Grady: I was going to say, is that a good thing or a bad thing? Because I think the positive case would be people are better informed, they’re just making better decisions, and they’re expediting a lot of the aspects of building a company that are not unique to their specific company. I think the negative case would be that people are blindly pattern matching and not understanding the causal determinants that actually lead to healthy businesses over time. Do you think it’s more the former, it’s a good thing, or more the latter, it’s a bad thing? Maybe somewhere in the middle, case dependent.

Manny Medina: To be frank, I haven’t stopped to think about that particular question. But if I were to put a qualifier to it, I think that the initial conditions are different, meaning they’re starting from a point in that I’m going to keep the company small, I’m going to own more of my destiny, so we’re going to raise in a different way. We’re not going to be growth at all cost. I’ve seen it from several founders already that they take more time honing in who their ideal customer profile is, in making sure that they scale that versus just scaling everything.

So when I started Outreach, I was always trying to figure out what are the edges of my market, who should I be selling to? And it turns out I can sell to everybody, and that actually makes the problem worse, right? Because I can sell to a startup, I can sell to a very large company—like, Adobe was one of my earliest customers. As much as I can sell to the Hillary campaign, as much as I can sell to a repo company. They all have communication problems, they all have workflow problems, but they all have completely different ways of getting to them, honing in the product for their particular solution.

And it was really hard for me to let go. And I sort of like wallowed in the sea of business that I had from every walk of life where if I were to relive my company, I would really hone it in with intention and say, like, “Okay, so what does make Outreach a billion-dollar run rate company?” And then start from there. I’m seeing these new founders coming with that point of view. I’m like, what is the narrowest profile that I can sell without a lot of friction, and then going up from there. Now the problem with that is that they may be restraining their ICP. You see what I mean? They may not be trying, they’re not trying to be experimental enough to grow the company from single product to multi product to platform.

Pat Grady: Yeah, but quality tends to scale, right? If you really nail it for a small audience …

Manny Medina: I agree.

Pat Grady: And you can start compounding from there, that’s probably better than being mediocre for a big audience.

Manny Medina: A hundred percent. And it just makes life more fun because your roadmap is a lot clearer. Your tickets, they all look the same. So the scaling of all the other operations is actually easier when you have one type of customer that you’re serving with excellence. And then you get known, the word of mouth gets bigger, all that happens more organically. Again, like, I wish I knew that when I started Outreach, and it was faster to like, all right, let’s narrow, narrow, narrow, narrow, narrow, and stay super focused on these two things or one thing.

Lauren Reeder: All right, Manny, so how does one get started with Paid?

Manny Medina: So it’s a great question. We are now onboarding manually to make sure that we have all the pieces dialed to your business. So you apply, and there is a short questionnaire, it’s like, you know, monetization, margins, and what space you are in. We’ll schedule a conversation for onboarding. And the beauty of being an early stage startup is that all the onboarding is done by me and my small team. We make sure that we’re capturing your agent work correctly. We’re making sure that we’re guiding you with best practices because we understand the market. At this point, we’ve seen pretty much everything under the sun, so we understand your market, how you should charge and give you a few ideas, get your invoicing trail going, get your margin trail going so you have visibility into how you’re making money. And we’ll be a short call away at all times. So just apply, make sure that you have an agent business and some customers, and then we’ll take care of you from there on.

Pat Grady: And what’s it like to be part of your team? Maybe what words would you use to describe the culture of Paid?

Manny Medina: I know this is super trite, but the first word that comes to mind is “fun.” And the reason it’s fun is because we’re serving an industry that in itself is fun. Everyone that I talk to that is building agents out there, they can’t believe they’re doing this for a living, you know? They can’t believe. They’re like, these are …

Pat Grady: What a time to be alive.

Manny Medina: What a time to be alive.

Lauren Reeder: [laughs]

Manny Medina: Like, they’re doing stuff that—and they’re getting paid to have fun. So, like, all our conversations are super fun. Like, I remember when I was in SaaS, there’s a lot of people who were stressed out and there was all this, like, you know, milestones and whatever. So maybe because everyone is early stage right now, so there is not a whole lot of growth rounds happening, so nobody has a number that they have to hit or like an efficiency they have to hit, but everyone is just enjoying the discovery. So it’s a little bit like that book from Bezos of, like, Invent and Wander. Everyone is innovating and wandering, like, living the art of the possible. With models improving every seven months, it’s like you have a new toy to play with every seven months. I think the ability—so this is understated. The ability to vibe code an idea, to just at least show how it works and then actually build it has made the iterations and the conversations so much faster. Like, you can just say, “We think we should build this,” or “I have a lot of energy around this thing,” and you can just write it, and you can see it work and you can tweak it a little bit and then present it in front of everybody and sell it like you’re selling internally to another person. So this ability for us to be brass tacks in terms of what are we building, for whom we’re building, being so small and getting so much work done in a small team and being so close to our customer because we’re early, it’s just fun. I think that’s the best word to describe it.

Lightning round

Pat Grady: Very cool. Love it. All right, let’s jump into the lightning round. You ready?

Manny Medina: All right, let’s do it.

Pat Grady: All right, question number one. Who is on your Mount Rushmore of founders?

Manny Medina: Jeff Bezos, for sure.

Pat Grady: Okay.

Manny Medina: He’s a big name founder in my life. I have a lot of friends that are founders. They really inspire me. Like Todd Olson from Pendo. He’s a great founder that—I just find he’s a great human being as well. I really like what Sam Altman has done with OpenAI. I think the ability to just push out innovation with regularity and be a showman in the world, I think it’s something that I aspire to be. You know, the Collison brothers are really special to me because they’re so voracious in their appetite and their reading and how they think about the world that, you know, they’re so—even though they’re very young, they’re so wise that I really look up to them.

Pat Grady: Awesome.

Lauren Reeder: What’s one piece of content that every AI founder should read?

Manny Medina: You know, I’ve been thinking about this one a lot because the thing that changed my perspective in AI is actually a very old book about a statistical natural language processing, and it was written by a guy called—his name is Chris Manning. It was mandatory reading at Stanford, computer science program on the natural language program course. I don’t know that it is anymore because it’s so old, but it’s like one of those, like, oldie but goodie kind of thing where they tell you how to—you know, how a Markov chain works, the length of the look back and look forward to try to predict the next word. It’s like the early days of machine learning. And a lot of the stuff that we’re doing right now is still based on this statistical approach.

Pat Grady: Yep.

Manny Medina: So I know that a lot of advances have been done, and we now cut tokens in many different ways and optimize in many different ways, but I feel like if you don’t understand where it all came from, I don’t know that you’re going to get where it’s going. So that book is actually not a hard read, even though it sounds like a hard read. It’s a fairly easy read. It’s written in a fairly normal and accessible way, and I can’t believe more people haven’t read it yet.

Pat Grady: What AI product can you not live without?

Manny Medina: Oh, Perplexity. That’s an easy one.

Pat Grady: What do you do with Perplexity?

Manny Medina: Oh, we do everything.

Pat Grady: [laughs]

Manny Medina: Like, it’s so weird. No, seriously, it’s so weird. Like, fashion advice or, like …

Pat Grady: [laughs]

Manny Medina: Or, like, you know—so the hardest thing for me is to have to take a driving test again. Like, come on! Like, you know, I know it’s the wrong side of the road and everything, but, like, having to relearn everything is just absurd. And I’ve been, you know, pushing away learning how to drive here in the UK, and my wife had to bite the bullet because one of us had to drive at some point. And she just went to Perplexity and looked for, like, you know, advice on finding a good person to help her how to drive. And, like, she found this super guy that has helped her go from zero to driving in, like, no time. Which I didn’t know Perplexity can do that, but, like, it can. So it’s taken over pretty much every aspect of our life. That, and of course, I don’t write anything without Anthropic anymore. Like, Claude is my constant companion for pretty much everything I do. It’s like a friend that I talk to and I shouldn’t be talking to.

Lauren Reeder: [laughs]

Manny Medina: If we were to have imaginary friends, Claude is mine.

Lauren Reeder: Models are commoditizing, yes or no?

Manny Medina: No, not yet. Not in the world of reasoning. I think that we’re just scratching the surface of where this can go. And there’s a lot to go. Like, if you see every new model coming out, the input tokens are far more expensive than the one they leave behind by a factor of six or seven or eight. Like, not by a small amount. And I think that this is just going to continue if we discover new things.

Pat Grady: On what date did we or will we reach AGI?

Manny Medina: I think it’s kind of here already, just in an underused kind of way. We haven’t really bottomed out everything a model can do. So I think it’s just behind door number three or something. Like, I think it’s already here, we just haven’t acknowledged that it is. Yeah.

Lauren Reeder: And then what’s your most optimistic future state for AI? Describe it.

Manny Medina: It’s a scaffold for human imagination. I think we also haven’t really absorbed how much smarter we get when we have somebody else doing a lot of the thinking and discovering for us. It’s sort of like when somebody props you up on the shoulders, it’s like all of a sudden you can see farther and longer. AI is doing the same thing for us. We will now be able to see farther and longer and come up with things that were impossible before. But such a time to be alive.

Pat Grady: Yeah. Last question. One piece of advice for AI founders.

Manny Medina: Stay focused on a very narrow set of customers. Don’t worry about TAM. Disregard—and they say this all over the world. Disregard VC advice about big TAMs. Small TAMs will be big TAMs as long as you deliver a superb experience.

Pat Grady: Awesome. Thank you, Manny.

Lauren Reeder: Thank you.

Manny Medina: Thank you. That was awesome.

Mentioned in this episode

Mentioned in this episode:

• The Hedgehog and the Fox: 1953 essay by philosopher Isaiah Berlin that divides writers into hedgehogs, that view the world through a single defining idea, and foxes, that draw on a wide variety of ideas
• Manny’s list of contemporary “hedgehog” companies in AI:
  • Quandri 
  • XBOW
  • HappyRobot 
  • Owl
  • Crosby
• CPQ: Configure, Price, Quote
• Invent and Wander: Book by Jeff Bezos and Walter Isaacson
• Foundations of Statistical Natural Language Processing: 1999 book by Chris Manning and Hinrich Schütze that Manny cites as a piece of AI content every AI founder should read. (still in print, companion site here)