Why I Pivoted StoryLearner Instead of Giving Up

I have a secret for you: the first version of storylearner.app looked pretty different!

It had a pretty compelling story generation feature, where you could influence the story, pick the characters, and so on…

It looked good — the illustrations were very compelling, and the stories were pretty fun. But I ripped it away. In this post, I’ll dive into how I realized I needed to make the change, what to change, and how I executed it.

The First Reality Check and Discouragement

In the beginning, I was riding the creative high of building something new. I had done a decent amount of research, and the space didn’t look too crowded. I felt motivated and inspired.

But over time, as I started talking to people and working on the product, the landscape began to shift.

I discovered tools like Language Reactor and started getting aggressively targeted by competitor ads — apps I hadn’t even known existed before.

It was like pulling at a thread: the more I looked, the more players I uncovered, especially once I started working on the mobile alpha. That’s when it hit me — building something around stories wasn’t as novel as I’d thought, and the market was crowded.

That realization was discouraging, but not all at once.

I’d alternate between moments of “oh no” and renewed optimism: yes, the market is competitive, but I have a unique angle, right? That sustained me for a while.

But the final blow came when I uncovered a wave of mobile apps doing very similar things (like Redle - Rapidly Learn Languages with News & Stories). It was harder than ever to believe I had a differentiated product.

I also became increasingly skeptical about the viability of a subscription model in such a saturated space — people are tired of subscriptions. I started seriously considering quitting.

Thought Process Behind the Pivot

Selling individual books had crossed my mind before, but combining subscriptions and per-product purchases seemed too complex for the initial version of the product.

When the discouraging signs piled up, I took a step back. I wasn’t afraid to walk away — I’d had fun, I’d learned a lot, I’d shipped something. But I also had a nagging feeling that there was still something worth salvaging.

At that point, I had standalone stories, story packs, and a working prototype of a book adaptation pipeline. That pipeline was actually a pain to build — adapting books well is hard. You need to simplify language to the right CEFR level, maintain good translations, generate quality illustrations, and chunk everything into manageable parts. But that difficulty made it valuable.

Books are inherently engaging, and the idea of breaking them into short, predictable chapters — inspired by both Duolingo and Brandon Sanderson’s writing style (I’m a huge fan!) — made them even more learner-friendly.

So I made a call: drop the generative story feature, focus fully on adapted books, and build the product around that. It felt like a tighter, more focused bet.

Books are easier to understand and sell: people already look for language books. They can try a few chapters and decide whether they want to pay — no long-term commitment. That’s easier for users, even if it’s potentially worse for recurring revenue. But let’s be real — if no one subscribes, there’s no revenue anyway.

Why the New Approach Is Better and More Promising

The new direction is more aligned with how people actually think. When someone hears, “You can read Sherlock Holmes in simplified Spanish, with sentence-by-sentence narration and illustrations, plus word lookups and review tools,” they get it. That’s an easy pitch.

Compare that to: “You can generate custom AI stories with predefined characters in various CEFR levels…” It’s more abstract, the stories seem arbitrary. It sounds like yet another language learning gimmick, a thing that you probably don’t need.

Generative stories were fun and they looked good — hiding that feature wasn’t easy. But in practice, the value of curated, thoughtful content is higher.

It has more practical engineering benefits too. It’s reusable, more stable, and cheaper to generate at scale. With LLMs being flaky in both quality and uptime, having static, high-quality content is a strategic advantage.

Books also resonate emotionally. There’s a sense of identity and aspiration tied to finishing a real book — even a simplified one. And a known title gives people a reason to care.

Additionally, there was a pretty big difference in positioning that unlocked a new distribution channel. A learning app competes with the traditional language school, but a book? A book would be complementary, it doesn’t threaten the traditional teaching model. It adds an additional way to monetize learning to people who are already paying for learning. One important lesson I learned in the past is that selling to people who don’t want to spend money can be very frustrating, it’s much easier to sell to people who are already spending on something similar.

Plan and Execution of the Pivot

I planned the pivot in three parallel tracks: content, app changes, and marketing.

App changes, surprisingly, were faster than I expected. I was productive and used Copilot and Claude to speed things up. I restructured the app to make books first-class entities. I added audio narration support, Stripe integration for purchases, a chapter-based reader experience, and a reworked UX with a catalog and library view.

Then, I adapted three full books. That part turned out to be trickier than expected. LLM outputs needed QA and lots of checkpointing, sentence splitting caused bugs in TTS generation, and with 800MB per book (mostly images), even uploads became a challenge - one of the books had more than 700 hundreds of illustrations, and I also generated per sentence narration files. I had to parallelize file uploads and add more automated checks to make the pipeline robust.

Marketing and launch prep are still in progress. I created a new landing page and started working on blog content, but the real push is still ahead. I am also talking to people in real world and the response seems to be enthusiastic.

Reflection

Looking back, I’m glad I pivoted and didn’t give up.

Storylearner is a small, but real product — focused, scoped, and clear in what it offers. It won’t be a unicorn, but it doesn’t have to be. I’m still proud of it. The build is done, but can be extended and improved. Costs are low. Risk is capped.

The next phase is about getting it in front of people. Success will depend on distribution and my marketing skills. But at least now I’m selling something people intuitively understand. It also doesn’t compete with language schools — it complements them — which opens up potential partnerships and new distribution channels.

I’m cautiously optimistic. I didn’t quit. I refocused. I’m not tied down, and I still have upside. Wish me luck!