The Vision
Built From the Life You're Already Living
Every story and episode you study from is generated from what you've actually done: the words you looked up, the articles you read, the conversations you had, built at exactly your level. Good enough that you'd want it even if it taught you nothing.
Why This Exists
Sophisticated learners already know how to close the plateau: sentence-mining. Hunt down new words and phrases in something you're actually reading or listening to, build your own cards, then go find native content worth spending your evening on. It works. It has always worked.
What it doesn't do is scale. Doing all three steps yourself, on top of a full day, is real, sustained work, not laziness, and almost nobody keeps it up long enough for it to matter. The plateau was never a knowledge problem. It's a friction problem: the method that works has never been something a normal person can sustain by hand. TraceSpark exists to close that gap, automating the combination rather than replacing the method that already works.
The Product, As It Stands
This isn't a preview of something bigger still being built. It's what you get today, at full maturity: three steps that already run end to end.
Save it without leaving what you're doing.
Long-press or share a word, phrase, or sentence from any app, in the language you're learning, and it's saved instantly. No new tab, no switching context, no interruption to what you were already reading.
FSRS-6 tracks what's fading.
Every capture is tracked by FSRS-6, the same open-source scheduler modern Anki ships. It knows exactly which words and phrases are about to slip, and exactly when, and brings them back inside your next chapter instead of a flashcard deck.
Your due vocabulary, written into the next chapter.
Stories and podcasts generated at your level, built to weave in the words and phrases you captured out in your own day, so review happens inside something you'd choose to read or listen to anyway.
How We Intend to Earn Trust, Not Just This Month
Generated content is easy to make once and hard to keep good. We hold content quality to a standing regression discipline, re-checked against fixed thresholds before any change to how a chapter gets written, rather than letting quality quietly drift the way mass-generated audio content often does elsewhere.
Personalization here isn't inferred quietly in the background. It's shown: because you captured this word last Tuesday, it's in tonight's chapter. You can trace every piece of content back to something you actually did, not a guess about who we think you are.
The other advantage compounds over years, not features: this is built and run by one person, in the open, responding directly to the people using it. A funded competitor with a support queue can't replicate that kind of presence at any price. It's slower to build than a feature list, and it's the part that's hardest to copy.
Built Around How You Actually Learn
Capture at work, listen on the way home.
Save a term from a report or an email during the day. By your commute, it's already woven into that evening's episode, no extra step in between.
The dialect you actually hear.
Stories and narration built around the specific regional accent and slang you're living around, not a neutral, textbook voice that doesn't match the street outside your door.
A series that keeps going.
Serialized narrative arcs that continue chapter to chapter, built to reward staying with the story, not standalone drills you abandon halfway through.
Continuing to build on this foundation.
I am building TraceSpark in the open alongside a small community of serious learners. Join them and secure your place.
