TL;DR: An infoproduct isn't one thing you sell once. It's three revenue layers: paid access, reviews/feedback, and premium content drops. A storefront can host the first and third. A storefront can sell "a review"; only a conversation can perform one. So the interesting question for a creator isn't "which store", it's which layers belong inside a chat assistant that also collects the money. This guide maps the three layers, gives you a rail-picker and a review-routing rule you can copy, and marks the one case where a bot is the wrong home entirely.
- Who this is for: creators, coaches and experts who already have an audience on Telegram and sell knowledge, whether that's courses, memberships, 1:1 feedback, or premium files.
- What you'll get: two copyable artifacts (which payment rail per item, and what the bot handles vs escalates), plus a clear line on where a bot loses to a course platform.
- Last updated: 2026-07-05.
The leak: you're selling three things and only charging for one
Consider the month after a cohort launches. The course itself sold fine, and the buyers are in. Then the real work starts, and none of it is metered:
- Someone DMs "can you look at my homework?" and you review it for free, at 11pm.
- Three people ask the same setup question, and you answer each by hand.
- You have a great bonus template you'd happily sell for $19, but standing up a checkout for one file feels like more work than it's worth, so you just don't.
That's three revenue layers leaking at once. The access sold; the reviews and the premium drops run on goodwill and manual effort. For most creators that recurring interaction is where the durable income lives, and it's exactly the part that never gets a price tag.
The reason is structural. Your course host (Kajabi, Teachable, Circle) is built to host and gate: video, drip, progress. It was never built to converse, so the conversational layer falls back to you, personally, for free.
The bundle: one assistant runs all three layers
The alternative is to let one Telegram assistant carry all three, and take the payment in the same chat instead of bouncing the buyer to a hosted page.
| Revenue layer | What the buyer gets | How the assistant handles it |
|---|---|---|
| Paid access | Entry to a gated premium tier: your locked content, guides, answer library | Collects payment, then unlocks and delivers the material in the chat |
| Reviews / feedback | A structured critique of their work, on demand | Does a first-pass review from your rubric, charges for the deep one, escalates the hard case to you |
| Premium drops | A single paid file: a template, a recording, a swipe pack | Sells it for Stars, delivers in chat, pays you out |
The through-line is the bounded thing that's genuinely hard to assemble elsewhere: an assistant that sells your content and invoices on your own payment systems, in one chat. Content sales run on Telegram Stars with USDT payout; bigger tickets run on the payment rails you already connected (cards, crypto). The conversation itself becomes the product, rather than a store sitting beside it.
Why reviews are the layer a storefront can't touch
Paid access and premium drops, a storefront can replicate: Whop, Gumroad and the rest gate files and take a checkout. The layer that breaks them is feedback, because a review isn't a product you unlock. It has to be performed, per person, per submission.
Here's the shape of it. It's 11pm, and a buyer sends a portfolio and asks "is this landing page any good?". The assistant reads it against the rubric you gave it and returns a structured first pass: three things working, two to fix, a next step. It offers the deep 1:1 teardown behind a paid invoice. If the submission is genuinely thorny, it doesn't bluff; it flags the case and hands it to you the next morning. It escalates by exception, not by default.
A storefront can sell "a review" as a SKU. It cannot do the review, decide when it's out of its depth, or charge for the version that's worth more. That's the specific job the conversation unlocks, and it isn't just a cheaper store: it's a different capability.
The routing that makes it work is simple enough to copy. This is the second artifact to steal for your own bot:
| What the buyer submits | What the assistant does | Money |
|---|---|---|
| A quick "is this okay?" that matches your rubric | Free or included first-pass review | No charge (it's your funnel) |
| A request for a deep, specific teardown | Runs the structured review, then invoices | Paid, in chat |
| High-stakes / legal / brand-sensitive / off-rubric | Flags it and routes to you | You quote it |
A boundary worth naming up front: the assistant does a first pass well and escalates the rest. Expect to skim its early reviews and correct its calibration before you trust it unwatched.
Which rail for which item
The one thing to get right is the payment rail, because it decides how much of each sale you keep. Telegram Stars are frictionless for an impulse buy, but Stars bought inside the mobile app carry Apple and Google's ~30% in-app-purchase cut, charged on the buyer's purchase. It's a buyer-side platform economics thing, not a Telegram withdrawal fee, and the real mechanics are worth reading before you price anything. For anything with a real ticket size, you invoice on your own connected rails instead, where the economics are governed by your processor's fee rather than the in-app store path.
So the rule is ticket-size shaped. Copy this table, fill the middle column with your own items, and you have your monetization plan:
| Your item (fill in) | Typical ticket | Rail | Why |
|---|---|---|---|
| e.g. bonus template pack | under ~$25 | Telegram Stars | Impulse, one-tap; the in-app cut is worth the zero friction |
| e.g. paid access tier | ~$25–75 | either | Depends on volume; run the fee math per your numbers |
| e.g. cohort review package | over ~$75 | Your own invoice (cards/crypto) | The in-app cut matters, so keep it on your rails |
A worked line, illustrative, so plug in your own numbers: on a $19 template pack sold for Stars in-app, the ~30% store cut on the buyer's purchase leaves your effective take closer to $13; on a $150 review package invoiced on your own rails, a ~3% processor fee leaves about $145 before tax, refunds and AI usage. The AI work itself (the replies, the first-pass reviews) is billed from a prepaid balance by the action, cents at a time, rather than as a slice of your sale. The full Stars-vs-own-rails math is in the fee guide; the step-by-step Stars → USDT payout is here.
The real cost, next to a platform subscription
The creator-SaaS stack charges you before you've earned. Rough 2026 numbers, research-dated, so confirm on each vendor's page before you decide:
| Option | What it costs (2026 ⚠️) | Hosts your course? |
|---|---|---|
| Kajabi | ~$143–399/mo subscription | Yes: video, drip, progress |
| Teachable | ~$29/mo + ~7.5% on the Starter tier | Yes |
| Circle | ~$89/mo + ~2% | Community + light hosting |
| Whop | ~3% platform + ~2.7% + $0.30 processing, per sale | Storefront + gating |
| Gumroad | 10% + $0.50 per sale | File delivery only |
| The assistant | Prepaid balance, pay per AI action; any take-rate on your sales is optional and isn't charged today | No: not a course host |
Read the last column, not just the price one: the assistant loses the hosting row outright. On money its bet is usage, not a take-rate — you pay for what it does (a reply, a review), and a percentage of your sales isn't charged today. That only pays off if the conversation is doing real work.
When a bot is the wrong home
If your product is a large one-time video course (forty lessons, drip, progress tracking, download protection) and there's little ongoing conversation after the sale, don't move it into a chat. A bot is not a learning-management system. There's no video player, no DRM, no course structure, no completion tracking. Keep the course on Kajabi, Teachable or GetCourse, where that machinery already exists.
The assistant earns its place when there's recurring conversation around the course: reviews, questions, a stream of premium drops. Below roughly a handful of paid interactions a week, the manual way is fine and a monthly subscription isn't. The clean pattern for a video seller is to run both: the course stays on its platform, and the assistant becomes the paid front door and the feedback engine around it.
FAQ
Do I need a separate store or website? No. The paid items and the checkout live inside the assistant and its Telegram Mini App. A big hosted course is the exception above.
How do I actually get paid out? Stars sales pay out to you in USDT, and you set the wallet once. Invoice-mode sales land directly on the payment systems you connected (your cards or crypto rails).
What happens when the assistant gets a review wrong? It's built to escalate, so genuinely hard cases go to you rather than getting a confident wrong answer. Plan to check its calibration in the first week, the way you'd onboard a new junior.
Isn't this just another storefront? The review layer is the difference: a storefront gates files, while this one performs feedback, charges for the deep version, and hands off what's beyond it. That's the part a store can't do.
Can I keep my existing payment provider? Yes. Invoice mode runs on the rails you already have; it doesn't replace them.
Next step
If you already sell knowledge and answer people for free in DMs, turn one of those answers into a paid item and watch what happens. New accounts get their first 30 messages free, so you can wire up an item, a price and a payout before spending anything. Start here: Turn on monetization. For the full picture across every model, see the complete guide to monetizing your AI assistant.
Sources & last updated
- Telegram, "Introducing Telegram Stars", telegram.org/blog/telegram-stars (2024-06-06). Stars in-app-purchase mechanic; Fragment payout.
- Telegram Stars effective fee (mobile ~30% in-app cut vs desktop/web): third-party creator-economy analyses; see the linked fee guide for the worked breakdown.
- Competitor pricing (Kajabi, Teachable, Circle, Whop, Gumroad): vendor pricing pages; research-dated, verify before relying on exact figures.
- iSales pricing/model: prepaid balance, pay per action, first 30 messages free; product pricing reference.
Competitor prices checked July 2026 — list prices, subject to change.
