CreatePie is live: a web platform for publishing courses, offering one-to-one sessions and scheduling webinars. We're sharing the product we're building, along with a few implementation choices that matter when learning content and payments meet.
The central business rule is simple: CreatePie charges for its software subscription and takes 0% sales commission. A creator's customer payment goes through the eligible payment account that creator connects. Provider processing fees and applicable taxes can still apply.
Separate the two payment relationships
A creator buys access to the software. A learner buys a course or books a session from that creator. These need separate identities, records and access rules.
| CreatePie plan | Monthly | Yearly | What it includes |
|---|---|---|---|
| Free | $0 | $0 | Publish free courses; no live scheduling |
| Courses | $19 | $199 | Free and paid courses, creator payment connections |
| Complete | $29 | $299 | Courses, one-to-one sessions, webinars and scheduling |
These are USD subscription prices. Yearly Courses saves $29 and yearly Complete saves $49 compared with twelve monthly payments.
CreatePie uses Dodo for its own subscriptions. Creators can connect their own eligible Stripe, Razorpay or Dodo account for customer sales, and multiple providers can stay connected. Creator Dodo sales are for eligible digital courses; live coaching and webinar payments require an eligible Stripe or Razorpay account under provider policy. The payment setup guide explains the distinction.
On the application side, checkout context belongs to the authenticated creator. A provider event must resolve to that server-side context before it changes the creator's plan. A successful browser redirect alone is insufficient evidence to unlock a subscription. Signed event handling and repeat-event protection are part of that flow; Dodo documents its webhook contract in its official webhook guide.
Put course access in front of file delivery
Creators can write native lessons and upload MP4 videos, PDF materials and project ZIP files. CreatePie uses Cloudflare Workers, D1 and R2 for the application, records and file storage.
The useful boundary is the application route serving a file. It checks the requesting creator or learner's course access and the asset's course association before reading the object. Successful private deliveries use private, no-store, carry noindex directives and support byte ranges for video playback. A file address is an identifier, not a bearer grant.
Cloudflare's R2 Workers API reference describes how bucket bindings expose object metadata and streams to a Worker. The authorization decision remains application code. Cache headers and crawler directives cannot replace it.
The live course audit downloaded all fourteen published assets and compared their hashes with the local originals. It also checked anonymous and unrelated-creator denial for representative video, PDF and ZIP requests, including HEAD, plus authorized range delivery. That is useful release evidence with a specific scope; it is not a blanket security certification.
A waiting room needs a clock and a booking
Complete includes scheduled sessions, seat limits and individual attendee waiting rooms. After a confirmed booking, an attendee receives room credentials by email. Another email arrives thirty minutes before the scheduled start.
An attendee can enter the room early and see the countdown. The meeting link is revealed during the scheduled session window. Once it is visible, the meeting provider's admission controls matter: a revealed meeting link can still be forwarded.
Confirmation and automatic-reminder delivery have been observed in a real inbox using seeded bookings. Real charged subscription settlement, renewal, refund and paid-learner fulfillment acceptance remain financial tests to complete; the release audit did not create a purchase or fake a paid entitlement.
A course you can inspect
The first original Computer Science to AI course is published for ₹5,000 as a one-time learner purchase, separate from creator subscriptions. It includes:
- Twenty-four complete lessons with worked examples, exercises and answers, covering CS foundations, Python, data structures, mathematics, OS, networking, SQL, machine learning and grounded AI.
- Twelve short narrated overview videos, totaling about thirty-six minutes, with synthetic narration disclosed.
- A forty-six-page black-and-white workbook and a runnable Python project with SQLite audit storage.
- An offline, extractive assistant with cited passages, a transparent ML experiment and evaluation exercises.
The estimated forty to fifty-five hours includes reading, exercises, labs and the capstone. Twelve existing short overview videos accompany their corresponding lessons; they do not cover all twenty-four lessons or represent the study duration. Two CS foundation lessons provide free text previews. Private uploaded files still require authorized course access.
You can inspect and reuse the free Python project source on GitHub. It includes synthetic-data machine learning, cited offline retrieval, HTTP and SQLite without an API key. The MIT project can be used independently of the accompanying course.
Original course prose and CreatePie-created media are CC BY 4.0; the original project code and synthetic data are MIT. External study references keep their own licenses. The hosted CreatePie application is proprietary. CC BY 4.0 permits commercial sharing and adaptation with its attribution conditions; the course's source ledger records each linked resource separately.
Make the public explanation useful
The original cake logo and pink design remain part of the product. Public course, creator and guide pages have rendered titles, descriptions and canonical URLs. Published pages enter the sitemap; private lessons, dashboards, rooms and checkout stay outside it. Checkout is marked noindex and points its canonical link to the public course page.
Clear pages help a person understand the product before signing in. They also give search and answer systems factual material to read, without promising a ranking or an AI recommendation.
You can start with a free course, compare the plans or inspect the published course above. We'd especially like feedback on the creator-to-learner handoff: what would make the course page, checkout or waiting room clearer for your students?
Top comments (1)
For the creator-to-learner handoff, I'd make "confirming payment and course access" a distinct state from both success and failure. Your server-side event boundary is useful, but a learner returning from checkout may arrive before the event has granted access.
A helpful acceptance case would be: delay that event, refresh the return page, then deliver it twice. The learner should see a clear pending message without being told to buy again, and fulfillment should happen once when the verified event arrives. Showing the course and an order reference in that pending state would also make a support request easier. This is a proposed test, not a claim about how CreatePie behaves today.