Santi: I had a client — this was maybe six months into the consultancy — who wanted a custom AI content pipeline. Totally bespoke. Different models for each stage, custom integrations with their CMS, a scoring layer I had to build from scratch. Took me three weeks.
Kira: Three weeks for one client.
Santi: Three weeks. And I was in Chiang Mai at the time, so half of those weeks were me working until two AM because the client was in Toronto and wanted synchronous check-ins.
Kira: Of course they did.
Santi: Delivered it. They loved it. Paid me forty-two hundred. And then the next client walks in and says, "I want something similar but for e-commerce." And I'm staring at my Notion board thinking... I just built something I can't reuse. Not a single template. Not one reusable component. Forty-two hundred dollars of work that dies with this one engagement.
Kira: And your effective hourly rate on that project was what?
Santi: I did the math later. Twenty-eight dollars an hour. In Chiang Mai, fine, you can survive on that. In Lisbon? That's the Bali Trap in real time.
Kira: Twenty-eight dollars an hour from a guy who used to bill three hundred at Spotify.
Santi: And the worst part — the worst part — is I had four more leads in the pipeline, and every single one of them needed something "a little different." So I'm looking at four more three-week custom builds, four more rounds of scope creep, four more clients who want synchronous calls in their time zone. And I realized — this doesn't scale. This isn't even a business. It's freelancing with extra steps.
Kira: That moment Santi just described — staring at a pipeline full of custom work you can't templatize — if that's where you are right now, this episode is your exit ramp. By the end, you'll have a productized services plan. One offer, strict boundaries, a ninety-day execution timeline with weekly pass-fail gates, and a clear path from service revenue to something you can actually sell or spin out.
Santi: No theory. Actual numbers from founders who made the switch, the exact KPIs to track every Monday, and the de-risk checklist that keeps your margins alive while you prove it works.
Kira: So the problem with custom AI services — and I say this as someone who ran a custom content agency for two years — is variance. Every engagement is a snowflake. Different scope, different tools, different client expectations. And when you're operating across time zones, variance is the thing that kills you.
Santi: It's not just the delivery variance. It's the margin variance. I tracked this obsessively. My custom consultancy projects ranged from sixty-two percent gross margin on a good month to nineteen percent on a bad one. Nineteen. Because one client needed a custom integration that ate forty hours I hadn't scoped.
Kira: And you can't price for that when every project is different.
Santi: You literally cannot. You're guessing. Every proposal is a guess.
Kira: Which is exactly what Brian Casel figured out when he launched Audience Ops back in twenty fifteen. He'd been doing freelance content strategy — custom proposals, custom scoping, the whole dance. And his move was radical for the time. He put pricing on the homepage. Fixed deliverables. Standardized packages. No proposals.
Santi: And it worked. Five-figure MRR within months. Grew to a team of twenty-five, served hundreds of clients, and eventually sold the business for a high-six-figure exit in twenty twenty-one.
Kira: From freelancer to exit in six years. And the key insight wasn't some brilliant marketing play — it was that the system was the product. Not Brian's personal expertise. The templates, the SOPs, the delivery cadence. That's what scaled.
Santi: Okay, but here's where I want to push on this — because the obvious counterargument is that productizing commoditizes you. You become interchangeable. And there's real evidence for that concern.
Kira: You're thinking about Design Pickle.
Santi: I'm thinking about Design Pickle. Russ Perry launched an unlimited design subscription — three hundred seventy dollars a month, one-to-two-day turnaround. Hit a million-dollar run rate in seventeen months. Three hundred fifty clients. Eighty-nine thousand MRR. Incredible growth.
Kira: But?
Santi: But he said something in an interview that should terrify anyone thinking about productizing. Quote — "sales and churn are at one to one." For every new client coming in, one was walking out.
Kira: One to one. So you're running just to stand still.
Santi: Running to stand still. And the reason — the scope was too loose. "Unlimited design" sounds great on a landing page, but it means different things to every client. Some want social graphics. Some want pitch decks. Some want brand identity work. The variance creeps back in through the front door.
Kira: So the lesson isn't "don't productize." The lesson is productize tighter. Narrower scope, harder boundaries, and — this is the important part — templates that embed into the client's actual workflow so leaving is painful.
Santi: Switching costs. That's Hamilton Helmer's framework from Seven Powers. And it's the difference between a productized service that churns at one-to-one and one that compounds.
Kira: Sam Shepler at Testimonial Hero figured this out. He took bespoke video production — on-site shoots, custom everything — and standardized it into packages. Remote capture, async editing, templated delivery. The company hit three-point-four million in annual revenue, and he automated over a million tasks in a single year through Zapier and internal tools.
Santi: And Shepler has said publicly that gross margin is the single most important line item for a productized service. Not revenue. Not client count. Gross margin. Because that's the number that tells you whether your templates and SOPs are actually working or whether you're just doing custom work with a fixed price tag.
Kira: Which is worse than custom work with a custom price tag, honestly.
Santi: Way worse. At least with custom pricing you can charge for the chaos.
Kira: Okay, so let's get concrete. You're an AI service founder. You've got three, maybe five clients. You're doing custom work. You want to productize. What does the first thirty days actually look like?
Santi: Days one through thirty are about one thing — picking the offer and proving someone will pay for it. Not building the perfect system. Not automating everything. Just — can three people pay you a fixed monthly price for a defined scope?
Kira: And the scope has to be narrow enough that you can deliver it with templates. Imagine you're doing AI-powered email sequences for B2B SaaS companies. That's your thing. Not "AI content" — AI email sequences for one type of client.
Santi: Right. And you write the one-pager. One page. Your promise with a number — "we deliver four email sequences per month with a forty-eight-hour turnaround." Who it's for. What's included. What's explicitly not included. Price. SLAs. Refund terms. Done.
Kira: The "not included" part is where most people choke. They're afraid to say no.
Santi: Because saying no feels like leaving money on the table.
Kira: It feels that way. But every "yes" to an out-of-scope request is margin you're donating. I learned this the hard way — I had a client ask me to "just quickly" build a landing page because we were already doing their content. Took my contractor fourteen hours. Ate the entire margin on that account for the month.
Santi: Fourteen hours of free work because you didn't have a scope boundary written down.
Kira: Exactly. So the one-pager isn't a sales document. It's a scope enforcement tool. You send it after every discovery call. You pin it in your client channel. When someone asks for something outside the boundary, you point to the page.
Santi: Days thirty-one through sixty — this is where you instrument. You're tracking five numbers every Monday. Gross margin. SLA hit rate. Time-to-value — how many days from kickoff to the first accepted deliverable. Utilization. And churn.
Kira: Why time-to-value specifically?
Santi: Because it's the leading indicator for churn. If a client doesn't see a result in the first seven days, the probability they cancel in month two goes through the roof. I don't have hard benchmarks for AI services specifically — that data doesn't exist yet in a reliable way — but every subscription business I've studied shows the same pattern. Fast first value, low churn. Slow first value, high churn.
Kira: So your target is — what, seven days to first accepted output?
Santi: Seven days or less. And if you're missing that for two weeks in a row, that's an automatic fail. You stop selling and fix the onboarding before you add another client.
Kira: That's the de-risk checklist in action. You set guardrails before you start — SLA thresholds, model spend ceilings, margin floors — and when you breach one, you pause and fix instead of pushing through.
Santi: The model spend ceiling is the one that catches AI service founders off guard. You're running GPT-4o or Claude for client work, and you think you know what it costs per request. But then one client sends you a forty-page document to process, and suddenly your API bill for that account triples in a week.
Kira: So you set a ceiling. Say, fifty dollars per client per month in model costs. And you set an alert at eighty percent of that ceiling. When it fires, you either optimize the workflow, switch to a cheaper model for that task, or have a conversation with the client about scope.
Santi: And you build the fallback model into your stack from day one. Not after the bill surprises you.
Kira: Days sixty-one through ninety. This is where it gets interesting. Your margins are proven — you've got three weeks of sixty percent or better gross margin. Your SLAs are holding. Churn is under control. Now what?
Santi: Now you raise the price or lock in annual contracts. And you start building moats deliberately.
Kira: Okay, walk me through the moat scorecard. Because I know you've been itching to bring Helmer into this.
Santi: I have a spreadsheet. Obviously. So Helmer's Seven Powers gives you categories — switching costs, process power, cornered resource, distribution, branding, scale economies, counter-positioning. For a productized AI service, the ones that matter most are switching costs, process power, and cornered data.
Kira: Switching costs first. How does a service business create switching costs?
Santi: Templates that live inside the client's tools. If your email sequences are built in their HubSpot instance, with your prompt chains wired into their workflows, ripping you out means rebuilding all of that. That's not a cancellation — that's a migration project.
Kira: And process power?
Santi: Your SOPs, your QA harness, your prompt library. The stuff that makes your output consistent even when you're delivering from a café in the Algarve with spotty wifi. A competitor can copy your one-pager. They can't copy two hundred hours of refined prompts and delivery checklists.
Kira: But here's my concern with the moat conversation. Most of us are solo operators or tiny teams. We're not accumulating millions of data points. We're not building network effects. Are these moats real at our scale, or are we just telling ourselves a nice story?
Santi: No, that's a fair challenge. And I think the honest answer is — at our scale, the moats are thinner. They're real, but they're not permanent. What they do is buy you time. Twelve months of switching costs buys you twelve months to build the next thing.
Kira: Which is the spinout.
Santi: Which is the spinout. And this is where the whole plan comes together. If you've been running a productized AI email service for ninety days, you've accumulated prompt libraries, delivery templates, QA checklists, and performance data across multiple clients. That's not just a service anymore — that's the foundation of a micro-SaaS or a template license.
Kira: Content Snare is the perfect example. They were a content agency that kept hitting the same bottleneck — collecting content from clients. So they built an internal tool to solve it. And then they realized the tool was more valuable than the agency.
Santi: Same pattern with RightMessage. Brennan Dunn was doing consulting and training on website personalization. Built a tool to automate what he was teaching. The consulting became the customer research for the product.
Kira: So the service isn't the end state. It's the laboratory.
Santi: It's the laboratory. But — and this is where I see people jump too early — you don't spin out until your margins are stable, your delivery is instrumented, and you've identified a specific data asset or workflow that has standalone value. Day ninety of the plan has a gate for this. You look at your moat scorecard, you look at your margin trend, and you make a deliberate decision — spin out, or keep compounding the service.
Kira: Not both.
Santi: Not both. I tried both. Lost four months and a lot of sleep.
Kira: And this is where the Lisbon Test matters. Whatever you build — the service, the spinout, the templates — does it work from a café with sketchy wifi? Can you kick off a client async, no live call required? Is there a fallback model if your primary API goes down? If any of those fail, you haven't built a nomad business. You've built a remote job with extra anxiety.
Santi: Remote job with extra anxiety. That's going on a t-shirt.
Kira: I'm serious though. Every item on the de-risk checklist and every line on the moat scorecard should pass the Lisbon Test. If it requires you to be online, in a specific time zone, on a specific call — it's a liability, not an asset.
Santi: So remember that forty-two-hundred-dollar custom build I was doing in Chiang Mai? Twenty-eight dollars an hour, no reusable templates, no path to anything?
Kira: The one that made you question your life choices at two AM.
Santi: That one. If I'd had this framework — one offer, one page, strict boundaries, weekly KPIs — I would have caught the margin bleed in week two instead of month four. The plan doesn't guarantee you'll build the next Content Snare. But it guarantees you'll know whether your service is actually working before you burn through a quarter pretending it is.
Kira: And the whole thing we walked through today — the one-pager template, the scope and boundaries matrix, the weekly KPI dashboard with pass-fail gates, the de-risk checklist, the moat scorecard with the Lisbon Test baked in — we put all of that into the Ninety-Day Productized Offer Pack. It's on the Resources page. Fill in the brackets, set your thresholds, and run your first Monday review this week.
Santi: One thing to do before Friday. Just one. Write the "not included" section of your one-pager. That's it. List five things you will not do for clients. Pin it somewhere you'll see it. Because that list is the difference between a business that scales and a freelance gig that eats your life.
Kira: That's the whole move. Start with the boundary.
Santi: See you Wednesday.
Kira: See you Wednesday.