How our apps teach

Characters that ARE the curriculum

Most educational apps put a friendly mascot on the screen, then deliver a lesson. We do something different: in many of our apps, the characters themselves are the lesson. Each one shows up again and again across the app, demonstrating a single idea — a chess move, a logic mistake, a spelling pattern, a geometry rule — so your kid learns the idea by following the character's story.

If that sounds familiar, it should. Story Time Chess teaches young children chess through a cast of characters. We borrow the term distributed narrative from the games literature and extend it across apps, and we build on intrinsic integration: the learning sits inside the thing you play. The difference here: what that approach does for one game, Spark & Anvil does across every app in the library — math, reading, science, logic, and more.

Free. No ads. No tracking. Nothing to sign up for.

Play in your browser — free, no sign-up See the evidence ↓
A friendly cartoon character showing a whole split into equal parts to a delighted child having a lightbulb moment, with a warm mentor character alongside.

Watch, don’t read

Watch Sir Pinwell do a pin

In GambitTales, Sir Pinwell has one job: pins. He never explains them — he does them. Step through one move and watch a knight get trapped.

Meet the board. Sir Pinwell’s bishop sits on f1; Black’s king is on e8. Press Next.

  1. Meet the board. Sir Pinwell’s bishop sits on f1; Black’s king is on e8. Press Next.
  2. Sir Pinwell’s bishop slides from f1 to b5. Check!
  3. Black blocks with the knight: b8 to c6.
  4. Look along the line: b5, c6, d7, then the Black king on e8.
  5. The knight on c6 wants to move — but if it does, the king is in check. It can’t. That’s a pin. Sir Pinwell never explained it. He did it.
  1. Lesson 3. A different board. Same Sir Pinwell. Both sides have castled — the kings are tucked away on g1 and g8.
  2. Sir Pinwell’s bishop is on c4, looking straight down the diagonal at Black’s king on g8.
  3. Only Black’s f7 pawn stands in the way: c4, d5, e6, f7, then the king on g8.
  4. The f7 pawn wants to step forward — but if it moves, the king is in check. It can’t even capture sideways. Same move, different board. Sir Pinwell just did it again.

Say it out loud

“Sir Pinwell pins it in place — it dare not move and bare the greater piece behind.”

Most apps

A mascot cheers while a worksheet asks “What is a pin?”

Ours

Sir Pinwell pins the knight. The move is the lesson.

Play in your browser — free, no sign-up

Why this works

Kids' brains are built for story. They'll remember the personalities, likes, dislikes and intentions of every animal in Frog and Toad longer than they'll remember a multiplication chart. Our approach borrows that strength: we turn the curriculum itself into a story with a recurring cast, so the abstract idea travels on a memorable character.

Stanley Greenspan gives many examples of the principle that we first learn from our emotional lives and then later generalize this understanding to a more abstract concept. We generalize from the personal context to a more generalized emotional meaning and then to abstract thinking itself. So much good teaching works from the personal to the general, from the individual example to the broader generalization, from our own emotion to those of other people — from the specific to the general. Stories resonate with us because we identify with the characters and relate to them from our own experience.

  • Recurring, not random. The same characters appear across many lessons. Your kid sees Sir Pinwell in lesson 1, again in lesson 3, again in lesson 7 — each time showing the same kind of emotional dilemma and making the same move in a new situation.
  • Behavior, not decoration. The character's actions are the rule. Sir Pinwell doesn't talk about pins. He does pins. Your kid learns the pattern by watching, not by being told.
  • Small casts, not crowds. Each app uses about a dozen characters, and never more than eighteen. Kids can keep that many in mind — like a Pixar movie ensemble.
  • Story carries the work. When kids care about a character, they remember what that character did. That memory is the curriculum.

The four pieces

What makes the method actually work

Researchers who study educational games found four things that make a story-driven learning app stick — and four things whose absence makes it fail. Our apps are designed to do all four at once. Here's what each piece does in plain language.

Try it

Take a piece away

All four pieces are on. Switch one off and watch how a story-app falls apart without it.

Repeats

The same characters return across many lessons, so the pattern gets seen again and again.

Without this piece: A cameo. The pattern never gets a second look.

Same thing

The character’s move IS the rule. The story and the lesson happen in the same place.

Without this piece: The story is just wrapping paper. The drill is still a drill.

Small cast

A small cast — about a dozen — you can actually get to know and remember.

Without this piece: Too many to know. Nothing — and no one — to care about.

Notices

The app watches what your kid needs and quietly adjusts — all on the device.

Without this piece: The same lesson for every kid, whether or not it landed.

Piece 1

A cast that REPEATS

The story lives across many lessons, not in one place. Sir Pinwell shows up in lesson 1, again in lesson 3, again in lesson 7 — each time doing the same kind of move in a new situation. Spreading the story across the lessons gives your kid more chances to recognize the pattern.

We borrow the term “distributed narrative” from the games literature — the story scattered across the whole game world, not stuck on the first screen — and extend it across apps.

Piece 2

Character + rule are the SAME thing

The character's actions ARE the lesson. Sir Pinwell doesn't talk about pins — he does pins. The story and the learning happen in the same place. The character is never a sugar coating around an unrelated drill.

Researchers call this “intrinsic integration”: games teach best when the learning is inside the thing you play, not bolted on (Habgood & Ainsworth, 2011). We extend the idea to characters.

Piece 3

A small cast you can KNOW

About a dozen characters per app, never more than eighteen. Each character has a personality, a portrait, a backstory, a voice. When your kid cares about a character — actually cares — they remember what that character did. That memory IS the curriculum.

Researchers call this “empathetic characters” — we keep casts small so kids can hold the whole ensemble in mind.

Piece 4

Apps that NOTICE your kid

Our apps watch what your kid actually needs and adjust. When a kid gets stuck, the right character shows up to help — gently, without making a big deal. When a kid is racing ahead, the next lesson stretches a little farther. Everything stays on your kid's device. Nothing leaves it.

Researchers call this “adaptive storytelling” — the piece that turns a good story into a teacher who actually meets your kid where they are.

The four-piece framework comes from a review of two decades of research on narrative in learning games (Naul & Liu, 2020). Most kid-edu apps do one or two pieces well. Ours are built to do all four.

The AI mentor

The mentor asks — it doesn't answer

In 2025, a study in PNAS gave nearly 1,000 students at one high school ChatGPT for math practice. They did better on the practice — then scored 17% worse on the later exam than students who had no AI at all, because the bot did the thinking. A second group used a version rebuilt to give hints, not answers; their exam scores showed no drop. Our mentor is built on that second model.

So in our apps your kid isn't asking a bot for the solution — your kid is the teacher, and a cast character is an eager novice who's still learning. Your kid explains the idea, spots the character's mistake, and helps it improve. Stanford and Vanderbilt researchers named this the protégé effect: kids who teach a character put in more effort, and in the original study lower-achieving students who taught the character caught up with higher achievers on the hardest questions. It isn't a soft promise — the mentor offers questions and fading hints but structurally cannot type the answer, never opens a free-text chat, and runs entirely on the device, so nothing your kid says leaves the iPad.

It's also why our approach was designed for the 2026 rules rather than scrambling to meet them. It aligns with the FTC's updated COPPA Rule (in effect since June 2025; full compliance required from April 22, 2026) and the direction of proposed laws like the federal GUARD Act (passed committee in April 2026; not yet law as of September 2026), California's SB 243, and New York's AI-companion law (both in force). There's no open-ended companion chat, no persona to bond with, and nothing that names you leaves the device — so the behaviors those rules target aren't there. Safe by design, not a setting you have to remember to switch on.

See it in action

Teach Sir Pinwell

You’re helping a novice see a pin. Pick what to say — the mentor only ever asks, and it can’t hand over the answer.

No typing here — the mentor takes choices only.

The bishop’s line runs through the knight to the queen behind it.

Sir Pinwell: This knight can move to e4 — nothing’s stopping it.

The mentor: What’s behind the knight?

Sir Pinwell: The queen! If I move, the bishop sees the queen.

The mentor: So — can the knight move?

Sir Pinwell: It can’t. Not while the others are watching. That’s… a pin.

Ask it to just tell, and it won’t: “I can’t hand over answers — I’m built that way. But I can ask: what’s behind the knight?”

Why it’s built this way

A 2025 PNAS study: students given an answer-giving bot scored 17% worse once it was taken away; a hint-only version erased the drop. Kids who teach a novice work harder and learn more — the protégé effect.

This web demo is scripted. In the apps the mentor runs on-device with Apple’s FoundationModels — no free-text chat, and nothing your kid says leaves the iPad.

Designed to clear the 2026 rules (COPPA amendments, GUARD Act direction, CA SB 243, NY’s companion law) — the line-by-line mapping is on /safety.

By age

The same method, tuned to your kid

Pick a band — the method is the same; the register, pace, and support change. Your choice syncs with the rest of the site.

Ages 3–5

What changes
Audio-first and low-text. A grown-up reads along; nothing needs reading to play.
The mentor
A gentle character who shows, never quizzes — big targets, calm pace.
See the apps for this age

Ages 6–8

What changes
Early-elementary reading, short prompts, lots of pictures. Read-aloud is one tap away.
The mentor
The mentor asks small, concrete questions and celebrates the try, not the streak.
See the apps for this age

Ages 9–14

What changes
The tween core: the full curriculum, with a Standard and an Advanced register you choose.
The mentor
The mentor withholds the answer and fades its hints as your kid gets it.
See the apps for this age

Ages 15–18

What changes
A mature, realistic register — case scenarios, not cartoons — in the older-teen teaching studio.
The mentor
A peer study partner: it questions your reasoning, never hands you the solution.
See the apps for this age

Meet the cast. Each one does one thing.

About a dozen characters per app — a world small enough to know. Here are twelve, one idea each.

Every portrait and audio drama here is made with AI tools directed by one person — that’s how a free, ad-free nonprofit can afford thousands of characters. Nothing here depicts a real person.

Meet the whole cast →

One character, seven lessons

“Recurring, not random” is the first principle above — so here it is, shown not told. This is the same knight, Sir Pinwell, across seven GambitTales lessons. Same patient hold; a new trap each time.

  1. Lesson 1 Pins a lone knight to the king — now it can’t move.
  2. Lesson 2 Pins a bishop, then quietly wins it.
  3. Lesson 3 Pins the piece guarding a checkmate square.
  4. Lesson 4 Holds a cross-pin — two pieces frozen at once.
  5. Lesson 5 Pins a piece to the queen, the bigger prize.
  6. Lesson 6 Keeps the pin while the rest of the attack arrives.
  7. Lesson 7 Same idea, a tougher board — and it still holds.

One face, seven frames. When your kid meets Sir Pinwell in lesson 7, they already know him — that recognition is the method working.

Emotion is where understanding begins

Reading depth

A developmental therapist who works with kids ages 9 to 14 recently read through our chapters and told us something we'd been working toward but hadn't yet named on this page: kids learn abstract ideas best when the ideas are connected to how the kid feels in the moment of meeting them.

That insight has a name. DIR/Floortime is a developmental approach to learning created in the 1980s by pediatric psychiatrist Stanley Greenspan, used by therapists and educators worldwide. Greenspan's starting point was a durable, simple idea: thinking always begins in feeling. A kid encounters an idea through their own emotional experience of it — curiosity, surprise, frustration, recognition — and that felt moment is what anchors the idea in memory long after the lesson ends.

Our chapters lean into this. Each character isn't only demonstrating a rule — they're feeling something while they demonstrate it. Sir Pinwell isn't a chess-mechanics tutorial; he's a knight who patiently holds positions in place, and that patient holding IS the pin tactic. The emotional posture and the rule arrive together. That pairing is what makes the rule stick.

We surface this at the page level too. Many of our chapter pages now end with a “Pause to think” callout — three or four short, italic prompts that invite your kid to notice their own reaction to the character. Not “what does the rule say” but “when in your life have you done what they did?” Those prompts aren't decoration. They're the moment where the abstract idea meets the kid's own feeling — which, in Greenspan's framework, is where learning actually happens.

For families, educators, or clinicians familiar with the framework: yes — this is FEDC Capacity 6 thinking (the “logical thinking and building bridges between ideas” capacity) meeting the chapter as a designed scaffold. The reflection prompts are the affect-anchor step; the recurring character carries the bridge.

The research behind it

This isn't a hunch. The approach combines well-established findings from educational and cognitive research:

  • Design constraint Emily Naul & Min Liu — the four-piece framework (2020). A 2020 review of 26 studies of narrative in learning games found four features that recur in games shown to work: a story spread across the lessons, characters whose actions ARE the rule, an empathetic cast you can remember, and a system that adapts to the learner. These are the four pieces above. Spark & Anvil apps are built to land all four together.
  • Adoption Adoption in the market — Story Time Chess (since 2011). Story Time Chess, founded as Chess at Three, teaches young children chess through a cast of recurring characters and has taken that character-first approach to over 1,000 US schools. That is adoption, not a learning study. Spark & Anvil's contribution is breadth: applying the same character-driven method across an entire library of subjects rather than a single game.
  • Mechanism Jerome Bruner — narrative-mode thinking (1986). Bruner showed that people have two distinct ways of making sense of the world: the logical-categorical mode (rules and definitions) and the narrative mode (stories and characters). Both teach, but formal subjects like math, grammar, and logic become much easier to grasp when bridged through narrative. Arguments convince us of their truth; stories, of their lifelikeness — which is why every story here ends by cashing the character out in the rule.
  • Mechanism Stanley Greenspan & Serena Wieder — emotion as the anchor of learning (DIR/Floortime, 1998). Greenspan's developmental framework, used by therapists and educators worldwide, holds that symbolic thinking grows out of emotional experience. Our recurring characters and our “Pause to think” prompts both operationalize this: an idea presented alongside how a character feels about it sticks; the same idea presented as a rule alone often doesn't. A practitioner working in this framework recently reviewed our chapters and named the connection between abstract concepts and emotional experience as the design choice that makes them work.
  • Mechanism Anna Sfard — two metaphors for learning (1998). Sfard warns against choosing only the acquisition view or only the participation view of learning. Our cast supplies the participation side: when Sir Pinwell IS the pin tactic, your kid joins a community of practitioners (real and fictional) who use the move. The apps' practice supplies the acquisition side.
  • Mechanism Green & Brock (2000); Mar et al. (2021) — narrative transportation and memory. Readers absorbed in a story change their attitudes more and counter-argue less (Green & Brock). Across 75+ samples, stories are remembered and understood better than expository text (Mar et al.). Stable, recurring characters invite the empathy and identification that drive that absorption.
  • Design constraint Habgood & Ainsworth — intrinsic integration (2011). In their Zombie Division study, children learned more, and chose to keep playing far longer, when the math was the game's core mechanic rather than a quiz bolted on. We extend that idea to characters: the character's behavior IS the rule, not a decoration around an unrelated drill.
  • Mechanism On-screen characters as teachers (meta-analyses 2013–2025). Studies of "pedagogical agents" (the formal term for on-screen characters in educational software) find small but consistent gains in learning and motivation. All of them test a single character; no study has yet tested a recurring multi-character cast, which is what we are building.

Each card is tagged — Mechanism (why it works) · Adoption (used widely; not a learning study) · Design constraint (what fails without it). Most of our evidence is mechanism — which is exactly why we don't claim proven test-score gains.

The four pillars of active learning (Hirsh-Pasek, 2015)

Education researchers use a simple four-pillar test for whether an app actually teaches (Hirsh-Pasek, Zosh et al., 2015). Here's how our method meets each one.

Active (minds-on)
Your kid predicts, chooses, and reasons — never taps to advance a video.
Engaged (not distracted)
One quiet stage; no confetti or streaks — the character is the point, not a reward.
Meaningful (connected)
The idea rides a character your kid already cares about, across many lessons.
Socially interactive
Your kid teaches a character who is still learning — the protégé effect.

If you're a teacher, parent, or researcher who wants the full bibliography or per-app curriculum mapping, email educators@sparkanvil.org.

Read the full bibliography →

What we don't claim yet

Our approach is grounded in published research on how children learn — but we haven't yet run our own controlled study measuring test-score gains. The evidence is strong on the mechanism (why stories and characters help ideas stick) and honest about the rest. We're a young library, and we'd rather show you what we're building than promise an outcome we haven't measured.

The rules we build by

Twenty-three design laws shape every app we make — honest dice, no streaks, a world that's a file you keep, a word that only counts once you mean it, a summary we audit against its source, and more.

See all 23 design laws →

The words we use, defined once →

555

of our 586 cataloged apps teach this way — characters that ARE the curriculum.

Last updated 2026-11-03 · numbers refresh every ship.

How the method maps across the library

Not every app teaches with a cast, and that's on purpose. Here's the whole library, split by how each app teaches — so “where it doesn't fit” is a visible slice, not a footnote.

How our apps teach, by share of the library

555 Characters ARE the curriculum
A recurring cast whose behaviour is the rule.
5 Open-ended by design
Creativity apps. Their cast stays off the canvas, at the edges of the studio.
26 Other teaching modes
Drills, simulations, tools — taught without a cast.

Across our 586 cataloged apps. The Distributed-Narrative family was the first set built cast-first; the method has spread across the library since. Last updated 2026-11-03 · numbers refresh every ship.

Where this lives today

Our Distributed-Narrative family is the first set of apps built this way from the start. Each one has its own cast where characters embody the patterns the app teaches — chess tactics, Go formations, music motifs, backgammon plays, bridge bidding partners.

GambitTales • guided by Captain Castle

Narrative-Driven Chess for Tweens — tactical patterns + piece movements as recurring character archetypes (10-character cast: tactics layer + world layer). Captain Castle rook mascot narrates a kingdom where Pin / Fork / Skewer / Discovered Attack / Double Attack / X-Ray live alongside King Pumble, King Sable, Pawn Patrol, Sienna & Bran, Trotter & Trundle, Gable & Garrett, Queen Vesper.
FractionForge • guided by Halver

Master fractions, decimals, and place value through virtual manipulatives, number line challenges, and real-world problem contexts — following the research-backed Concrete-Representational-Abstract instructional model. Grades 4–6 (about ages 9–12).
ReadQuest • guided by Margin

Build reading comprehension mastery through interactive passage analysis, evidence-based argument construction, and AI-generated comprehension challenges — with dual fiction and nonfiction tracks. Ages 9-14.
CuriosityQuest • guided by Lumen

A daily-question adventure that builds curiosity into habit.

FractionForge in one line: a fraction is a plan for sharing, and a good plan has no sliver. Try the Orange Table →

Browse all 622 apps → Play in your browser →
Recently shipped

We're extending this teaching style across the portfolio, app by app, wherever it fits. Recently shipped in the Logic cluster:

  • LogicQuest Now live

    A dozen friendly-but-flawed arguers (each making a different classic mistake) plus four logic detectives who teach kids to spot bad reasoning by example.

  • EscapeForge Now live

    Seven puzzle-archetype characters — one each for math, word, cipher, pattern, logic, connection, and sequence puzzles — where each character IS the kind of puzzle they stand for.

  • DebateForge Now live

    A five-member debate team — Build, Weigh, Steel, Reply, and Yield — who each act out one move of a good argument: making a case, weighing evidence, steelmanning the other side, responding, and conceding gracefully.

Where the cast steps back

Some apps are about your kid's own creativity: BeatForge for making music, SpectrumCanvas for drawing, PixelForge for pixel art, SoundSphere for sound design, SynaForge for combining the senses. Their casts stay at the edges, in the stories around the studio or as one line when a tool is picked. A character never takes over the canvas: nobody draws for your kid, judges the work or says what to make. What matters most there is your kid's own ideas.

We also keep the casts small, diverse, and reviewed for representation before any character ships. A storytelling approach amplifies whoever's on stage, so we're careful about who we put there.

See it for yourself.

Play in your browser — free, no sign-up Support the mission

Spark & Anvil is a nonprofit; our 501(c)(3) application is pending.

See all 23 design laws →

The words we use, defined once →

For parents

You don't need to know any of this for the apps to work. Your kid just plays — they'll meet the characters naturally, follow their stories, and pick up the patterns without anyone telling them they're learning. The method is built in. That's the point.