--- title: "Course platforms, AI, and the retention gap" canonical: "https://filipsardi.com/course-platforms-ai-retention/" pubDate: "2026-03-09T00:00:00+00:00" author: Filip Sardi description: "Filip reviewed Skool, Circle, GoHighLevel, Teachable, Thinkific, and Kajabi with two questions - what are they building with AI, and how does it actually help clients get results." --- Can't believe we're already on our 20th consecutive weekly letter. I wanted to make this a special issue and decided to cover: - Two retention problems - same word with different outcome - Why client behavior inside paid programs has changed, and what that means for your delivery - Three coaching platforms that shut down in 2025, and what it revealed - Skool, Circle, GHL, Teachable, Thinkific, Kajabi: what they're actually building with AI right now - The gap no one's building into yet - What to build to get ahead in your market --- There's a question that used to light up every coaching Facebook group like nothing else. > "I'm thinking of switching platforms. Which one would you recommend?" A hundred comments with strong opinions and platform tribes competing for status. And I was always in there with a confident, specific answer - Kajabi for this, GHL if you want control, Circle if community is your model. That version of me stopped showing up a while ago. ### At some point I noticed that the louder a platform tribe gets, the more it becomes the trendy go-to, regardless of whether it actually helps clients get results. People switch, the interface changes, but the underlying problems stay the same. --- ## 1/ Two retention problems The problem isn't which platform to choose. It's which retention problem you're actually solving. ### You're trying to keep your clients engaged and getting results, while platforms are trying to keep you paying. Both call it retention. The motivation behind each one couldn't be more different and most platforms are optimizing for theirs, not yours. **Just to be clear about my lens going into this deep dive:** my main interest is performance - platforms having features that actually help clients make progress. If your goal is clients who engage, vibe, and have fun in community, and that reliably drives your retention metrics - that's legitimate, it's just not what I'm solving for here. --- ## 2/ What changed in the last two years Client behavior has changed more in the last two years than in the previous ten, and you've probably felt it already. If your clients are saying "I'm behind" more often than "here's what I did" - here is why: **Clients don't finish long videos anymore.** Attention inside paid programs has dropped in the same way it's dropped everywhere else - not because your content is worse, but because the environment around it is noisier. **Every piece of content you add to your offers increases cognitive load before it adds value**. Your clients are already using AI to implement - asking it to summarize your modules, create action plans from your frameworks, troubleshoot their decisions. That's happening whether you designed for it or not. **And for the first time, it's technically possible, and easy, to detect language shifts,** engagement pattern changes, milestone gaps - and surface them to you before the client even knows they're stuck. That capability wasn't easily accessible two years ago. --- ## 3/ Dedicated coaching platforms shutting down This was one of the most interesting discoveries and even though most people in my online space aren't using dedicated coaching platforms, there's a clear pattern unfolding here worth paying attention to. Three coaching platforms closed or wound down in 2025. Nudge Coach. Practice.do. Profi.io. Different products targeting different audiences, but the same underlying pattern. All venture-backed, all positioned as the comprehensive solution for coaches, all unable to raise again when the capital market shifted and AI companies started absorbing most of the available investment. **Practice.do is worth pausing on.** They shut the doors on November 3rd. On paper, that's a date. In real life, a lot of coaches experienced it as a scramble - logins failing, assets suddenly becoming unavailable and users racing to export what they could. **Profi.io went further.** During the wind-down, a lot of coaches discovered there wasn't a clean export path - no obvious "download everything" button or simple migration file you could rely on. Many users later reported that export feature never existed. One platform gave you short notice and a narrow window and another gave you nothing. These weren't just business failures. They were the clearest possible demonstration of the two retention problems playing out at the same time. ### The platforms were focused on keeping their own metrics looking healthy on paper for investors (monthly active users, engagement rates, renewal numbers) right up until the doors closed. Your clients' momentum was never part of the equation. Remember, my goal isn't to scare you off platforms - it's to make sure your business lives in your systems, not their servers. --- ## 4/ What six popular platforms are actually building right now I'm not currently using any of these for my projects, but almost all of my clients are so I watch them closely. Here's what I found when I went through each one with two questions: - what are they building with AI? - what does their client delivery infrastructure actually look like? **At the time of writing this (March 2026), here is what's currently happening across the six platforms most coaches in my world are using.** --- ### Is Skool good for client delivery and results tracking? Sam Ovens (Skool founder) keeps the platform minimal by design focusing on: community feed, classroom, calendar, leaderboards. No AI features of any kind have shipped for now. The gamification measures engagement: posting, commenting, liking, video completion. In their own words, they use "progress tracking" - but this means lesson completion checkmarks, not implementation tracking A member can sit at the top of the leaderboard without implementing a single thing from the course. In my eyes - great if you are focused on pure engagement, not so great if you're focused on performance outcomes. --- ### What does Circle's AI actually do and what can't it see? Circle describes their AI Agents as "24/7 support, coaching, and guidance for members." They've shipped more AI than anyone else in the community platform category - up to 10 agents per community, trained on your content, with customizable personalities and an AI Inbox where you can read every conversation (available only to the highest monthly tier level that's around $400 per month at the time of writing) The example they show: *"What brings you to Clarity?" "Want help finding your first event or course?" "How do I connect with members in my field?"* It's genuine AI investment, and it's still pointed at community navigation and engagement. The agents answer when asked, help members find things, but they don't surface who's stuck, who completed a module but skipped the implementation exercise, or who's been quiet for three weeks. The "inactive members" metric is the closest they get to a delivery signal, but it's a binary (active/inactive), not a pattern or a momentum read. That's why whether your clients are actually moving forward towards the outcome or not is still not something the platform shows you. --- ### What is GoHighLevel building with AI in 2026? GHL positions itself as the true all-in-one, and their AI investment is the most aggressive in this group. They aren't messing around and their March 2026 changelog includes Conversation AI V3 with a visual flow-based bot builder, AI Voice Agents for inbound call handling and lead qualification, AI Builder for workflow automation, and Meta Ad Manager integration. Also, coming soon: AI-Generated Flows, Voice AI Nodes, Rich Communication Services. Their own language about Conversation AI V3: *"Capturing leads, nurturing them, automating workflows."* The membership and courses area exists, but recent updates there are minimal - a few engagement features, nothing approaching delivery intelligence. Every AI feature GHL has built is pointed at the same thing: getting more clients in the door (and these analytics are detailed). What happens to those clients once they're inside is largely not a GHL problem. --- ### What does Teachable's AI actually do for client results? Teachable's positioning has shifted toward something worth noting: > "In a world where anyone can ask AI for information, we're the home for those who educate with purpose, modernity, and humanity. Teachable powers human-led learning that drives student trust, connection, and results." They're positioning around results. Their AI Hub covers curriculum generation, quiz creation, subtitles, and translation - all creation tools, all pointed at helping you build courses faster. "Last Activity" is the closest Teachable gets to a momentum signal and it's a timestamp, not a pattern. Also, the progress percentage is calculated as: lessons marked complete + quizzes completed + videos watched to 90%. **This is consumption tracking dressed up as progress tracking.** I'll come back to this pattern, because it shows up again. --- ### What is Thinkific Thinker and does it track client progress? Thinkific launched Thinker on February 24, 2026 - an AI Teaching Assistant built directly into the learning experience, trained on the creator's content. It answers learner questions in real-time, explains concepts, helps find specific lessons, summarizes key takeaways, quizzes learners to check understanding, and supports problem-solving using the creator's frameworks. It's inside the learning experience itself, not just the community layer around it. Their positioning language: *"In 2026, AI has made content abundant, but content alone is no longer enough. The real differentiator is now the experience around it."* They're saying something real. Thinker answers when a learner asks, but it doesn't tell you which client hasn't logged in for two weeks, which one completed Module 3 but never did the implementation exercise, or which ones are heading toward a cancellation they haven't mentioned yet. Even their "insights" promise is framed around "informing new offers and revenue streams", focused on upsell signals, not implementation signals. Cohort reports let's you compare January cohort vs. February cohort completion rates. This is useful for course optimization, not client delivery. It answers "did my course redesign improve completion?" not "who is this specific client and what do they need right now?" Thinkific is the closest any platform has come to recognizing the real gap. The product isn't there yet. --- ### What is Kajabi Cofounder and what is it built for? Cofounder is the most sophisticated AI product in this space right now - it analyzes your funnel performance, recommends pricing strategy, and generates marketing copy in your voice with live access to your account data. The Kajabi development roadmap (they call it Timberline) also includes Kajabi Amplify (audience growth), Contact Lifecycle Tracking (lead → buyer → repeat customer), and multiple order bumps at checkout. Contact Lifecycle Tracking is worth noting specifically: it shows where every contact stands in their journey. What someone bought, when they became a customer, whether they've purchased again. Not whether they implemented anything or if they're making progress. The lifecycle it tracks is a purchase lifecycle, not a transformation lifecycle. Knowing what someone bought tells you nothing about whether buying it changed anything. --- ## 5/ The gap nobody's building into (yet) After going through all six, the pattern is the same everywhere. The spectrum runs from no AI at all (Skool) to slightly more sophisticated AI (Kajabi's Cofounder, Circle's Agents, Thinkific's Thinker), but every AI investment is pointed at one of four things: - helping you sell more - helping you manage communities more efficiently - helping you improve completion rates - or helping learners find content when they ask for it What's interesting is that two of these platforms (Teachable and Thinkific) are using results-oriented language explicitly: > "Drives student trust, connection, and results." > "Content alone is no longer enough." They see the gap and they're naming it in their positioning, but the product doesn't close it. ### This isn't two companies making the same oversight. It's the industry's current relationship with the word "results" - easy to put in the positioning, structurally absent from the product. Every platform in this space exports the same six fields: Name, Email, Progress %, Logins, Start Date, Last Activity. That's not a delivery intelligence system, but a spreadsheet. Not one platform has shipped anything that proactively surfaces implementation signals to you as the creator. None of them can answer the questions that actually predict whether a client stays: - Who is executing between sessions, and who has gone quiet? - Who completed the content but skipped the implementation step? - Who is three weeks into a silence pattern that looks exactly like someone about to cancel without saying anything? **That entire category (proactive delivery intelligence) doesn't exist in any of their current roadmaps.** And the bar isn't even that high. I'm not talking about sophisticated AI systems or complex behavioral analysis. I'm talking about basic signals: who's implementing, who's gone quiet, who's drifting before they know they're drifting. The first one that does will have a significant advantage. Until then, that gap is exactly what I've been building into. --- ## 6/ What I'm building to get ahead My approach goes further than platform features, because it has to. **It starts at AI assisted onboarding, before the first call.** I use an AI tool to build a picture of where a client actually is: what they're carrying into the program, where they're stuck, what they're not saying yet. By the time we're on a call together, I already have context. That changes everything about how the first conversation goes. **From there, any teaching I do is short and pointed.** The goal is to hand off to implementation as fast as possible - not education for its own sake, but education that creates a clear next action. The videos my clients and I create are built around that- Specific and immediately actionable. **The implementation tools follow the same logic. They're not there to make clients feel productive** - they're there to make sure something actually gets finished. When a client works through one of these tools, the output is a set of specific tasks they assign to themselves. I can see what they committed to, and they know I can see it. That changes the quality of what happens between sessions. **And underneath all of it is Client Flow Pulse** - a simple weekly read on what's actually happening with each client, in their own words. Who's moving, who's stuck but not saying so and who needs something different from what I planned. The patterns that surface there aren't things any dashboard would catch, but they're the ones that determine whether someone renews or quietly disappears. ### None of this requires a different platform. It can sit inside whatever you're already using. --- Content delivery is becoming less important as AI evolves, while transformation, accountability, and coaching delivery are becoming way more important. ### The platforms haven't caught up with that yet, and the question is whether you will. So if this letter made you look at your own delivery differently (the platform you're on, what it shows you, what it doesn't) that's exactly where the Retention Multiplier Gameplan starts. We go through your client flow milestone by milestone. What's working, where momentum breaks, what your current setup can't see. And we figure out what to build inside what you already have before the gaps show up as churn. ### It's not a platform audit. It's a delivery audit. Reply with "GAMEPLAN" (or send me a message) and tell me what you're currently using and what you think your biggest drop-off point is. That's enough to start. *-Filip "writing to you 20th week in a row" Sardi*