Configure it, or own it.
EdTech software development.
Flashside, our own flashcard app, started in April 2020 and runs FSRS scheduling on one React Native 0.86 codebase and a .NET 10 API. In September 2026 a 114 MB Anki package had to import whole, and Timeplanned, our other product, turns tutors' attendance into month-end billing: education software development we did for ourselves, so the trade-offs below are ones we paid for.
A packaged learning app answers which deck, not what to study
Flashside spent its first six years answering one question: which deck do you want to open? The question a learner actually needs answered is what should I study today, and that gap is where edtech software development earns its cost, because a packaged app's scheduler, import and storage fit most learners most of the time.
Without that answer, a learner with 400 cards re-drills cards known cold for a month while the cards on the edge of forgetting wait their turn. That was Flashside's own position for its first six years, which is why the rows below come from our products and not a client's.
Four things we built past the box
Both products here are ours, so both columns are: what the packaged version of each feature does, and what edtech software development turned out to mean once real learners and real decks arrived.
| What a packaged tool handles | What we built past it, in our own products | Where we built it |
|---|---|---|
| What a packaged tool handlesA scheduler with FSRS or SM-2 already preconfigured. | What we built past it, in our own productsFSRS with a target-retention dial instead of SM-2's fixed ease factors. It is pure arithmetic and runs with the network off, and each grade button shows when the card returns, from under a minute to 14 days on a fresh card. | Where we built itFlashside |
| What a packaged tool handlesImporting a CSV or a small .apkg file. | What we built past it, in our own productsIn September 2026 the import was tested against Kaishi 1.5k, an Anki package of 1,501 notes with 14 fields each, 2,971 audio clips and 114 MB of media, which had to import whole, on an app and an API that knew nothing about audio, note types or Anki before that. | Where we built itFlashside |
| What a packaged tool handlesCloud-first storage with an account to sign in. | What we built past it, in our own productsCards and study history stay on the device, so the app works without a connection and without an account; the server is a meeting point for sync and sharing, not the system of record, so a lost connection loses nothing. The review log sits in SQLite on the device, one row per graded answer, 400 days kept. | Where we built itFlashside |
| What a packaged tool handlesAttendance tracking and an invoice template. | What we built past it, in our own productsRecorded attendance becomes month-end billing automatically, and the month-end amount accounts for cancellations, makeup lessons and one-off payments without a manual step. | Where we built itTimeplanned |
Six years of answering the wrong question
Flashside, an offline-first flashcard app
Our own product, started April 2020
Flashside started in April 2020 as a flashcard app, and three decisions made early shaped everything after. Cards and study history stay on the device. The server is a meeting point for sync and sharing, not the system of record. And whatever the AI does, remembering stays the learner's job.
Underneath are one React Native 0.86 codebase without Expo for iOS and Android, a .NET 10 API on MySQL 8 and a Next.js 16 web app. Sync sends deltas per collection and moves media four files at a time with retries; on iOS the upload runs on a background NSURLSession, which keeps going after the app is suspended.
The Anki import is where the plan met a real deck. Kaishi 1.5k carries 1,501 notes of 14 fields each and 2,971 audio clips, and the cards had to move out of key-value storage into a file per collection, because Android's SQLite cursor refuses a single row over 2 MB, and Kaishi's text alone is 1.3 MB.
What does a tutor need from software at month end?
Our own product
Timeplanned, scheduling and billing for private instructors
Our own product: scheduling and billing for private teachers and tutors, designed and built end to end.
Scheduling is the easy half. The month-end amount has to account for cancellations, makeup lessons and one-off payments, and recorded attendance turns into that bill without a manual step.
Per-student consent records with one-click export and erasure, including minor students with guardians.
When a better-configured platform wins
- Your courses already run on the LMS you have.
- Spend the budget on content instead. Edtech software development pays where the scheduler, the import or the offline behavior is the product itself.
- You need an LMS for a whole university next term.
- Our receipts here are two products of our own, a flashcard app and a tutoring tool. For a campus-wide rollout tied into student records, ask the vendors who list one.
- You want the AI to answer for the learner.
- Flashside's AI builds a study plan, writes quizzes from your own cards and rewrites the ones you keep failing, but remembering stays the learner's job. If the product is an AI that does the homework, we are not the team for it.
- Your learners are always online.
- Then most of the offline-first work on this page is cost you do not need. A server-first design is simpler to build and fine for a classroom with reliable Wi-Fi.
Questions from people building learning products
Why are the receipts on this page your own products?
Who owns the code and the learner data if you build our product?
What happens when an exam is three weeks away?
How do you handle student data under GDPR?
What is Flashside built with?
The import test takes one afternoon
The fastest way to find the line is one real deck. Describe the import your tool refuses, or the study flow it cannot model, and thirty minutes with the founder is enough to tell whether the gap is a build or a configuration job.
Or describe it in the form, and expect a reply inside 48 hours.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.