EmbarkGaps & next
Building the 30-second pitch showed where the concept, the design and the build don't yet match the story. High means it undercuts the pitch or carries real risk.
What the 30-second story exposed
Writing the pitch forced the concept into four beats: a small step, Basti walks, you are not alone, and the world grows. Three of the four are strong in the app. The weakest is the one the pitch leans on hardest: company.
HighBody doubling is the emotional core, and the thinnest part at launch
What I found. On a solo journey the road says "You and Basti are on your way." Lanterns on the lake only appear with other ferry passengers. An empty ferry says "Nobody else boarded this one." Real-company-only is the right rule (ADR-0085 §7), but with no players there's almost nothing to see.
Why it matters. The pitch promises "every lantern is someone starting something right now". Day one has no one, which is the cold-start problem every social app has, and body doubling depends on it.
Proposed refinement. Add an honest, always-on aggregate from the CloudKit public database: "214 journeys started in the last hour", counted, never invented, shown on the road and the lake. Seed the first months with the TestFlight cohort and friend invites. Give Basti more road life (small vignettes, residents) so the solo case still feels accompanied.
HighThe company is visual, but the instruction is "put the phone down"
What I found. While you work, the phone is face-down. Right now the only sound is footsteps during the 3-second departure animation. The road itself is silent, and only the Watch taps footsteps.
Why it matters. Body doubling works through ambient awareness. A screen you have put away can't provide it, so the channel has to be sound and touch.
Proposed refinement. An optional, quiet road soundscape (lake, footsteps, a distant bell when another passenger boards your ferry), the Watch footstep haptics as the default for Watch owners, and a soft arrival chime. All behind the existing Sound toggle.
High"He appears on your real desk" rarely happens
What I found. AR needs the phone upright with the camera on the desk. In every device run the phone was lying flat, so Basti waited on the painted pier. That's by design, but it means the hero moment is the exception.
Why it matters. The pitch's second beat is built on AR. Most people lay the phone down flat, which is also what the app asks them to do.
Proposed refinement. Make the painted station the canonical start and AR an extra: when the phone is flat, a one-line hint "Lift your phone to see him on your desk". Or move AR to the departure moment, when the person is holding the phone anyway.
Medium"Not a timer", but the mechanics are 5 and 25 minutes
What I found. The two ways are Pomodoro lengths. Dial is the family's focus instrument (ADR-0024), so the overlap is visible.
Why it matters. If Embark reads as a timer with a cat, it competes with Dial and with every focus app.
Proposed refinement. Lead every surface with start, company and world, never the clock. Consider a 2-minute "just open it" way (initiation, not focus) and an open "until I'm back" way.
Design
Things the pitch had to leave out because they would slow it down. That's a signal they slow down the person starting, too.
MediumToo much on the screen at the moment of starting
What I found. The first-run home already shows streak 0, brass 0, rank 1, sound, settings, a ferry chip, suggestions and an order. Behind them sit about twelve systems: XP, brass, streak, spares, orders, chest, finds, patches, story, bond, ranks, season pass.
Why it matters. For the people Embark is for, every extra element on the start screen is friction at exactly the wrong moment.
Proposed refinement. Progressive disclosure: the first journey shows only Basti, the ticket and Begin. The HUD appears after journey 1, orders on day 2, ferries after journey 3, and Camp on day 3.
MediumThe world is a list, not one world
What I found. World shows six cards, and each district painting reuses a road painting from the journey.
Why it matters. "Your work paints his world" lands hardest as one continuous place you watch fill in, which the pitch had to fake with a single painting.
Proposed refinement. One painted lake map (a generated master plus region masks) with the six districts as areas that paint in, and the far shore across the water.
MediumAll three answers pay, including "Not this time"
What I found. Intended (ADR-0085 §2): starting is the win. But a 5-minute idle journey answered "Not this time" still earns XP, brass, a find and a patch.
Why it matters. It can feel hollow, and it can be farmed beyond the 60-second guard.
Proposed refinement. Keep it, but make the gap visible: "I did it" gets the richer cascade, and "I got going" gets an immediate one-more-leg offer at a lower cost.
LowKeep walking earns nothing
What I found. People in flow who keep going past the way get no credit.
Why it matters. It punishes the best outcome, sustained focus.
Proposed refinement. Count Keep-walking minutes toward XP and patches at a reduced rate, capped.
Execution
Verified facts from the code, the device runs and App Store Connect.
HighThe body-doubling path has never run end to end
What I found. Ferry matchmaking between two Apple Accounts, the SharePlay walk and the community lighthouse with real participants are all untested.
Why it matters. These are the features the pitch sells.
Proposed refinement. A two-device session: same ferry, two accounts. Then one SharePlay walk over FaceTime.
HighThe season pass sells a reward that doesn't exist yet
What I found. Camp shows Enzo as "Coming soon", but the premium track promises him at tier 20.
Why it matters. That's a consumer-trust and App Review risk: a paid track with an undeliverable item.
Proposed refinement. Swap tier 20 for something that ships, or hold the pass until Enzo and the city route exist.
MediumHardware checks still open
What I found. AR start, widgets, the Lock Screen card through a whole journey, Watch pairing and VoiceOver.
Why it matters. Each is a promise in the pitch's "everywhere you glance" beat.
Proposed refinement. Your checklist on the Status page.
MediumContent runway is a few weeks
What I found. 20 finds, 8 story parts, 36 world pieces plus the far shore. At about 2 patches a journey, a daily player paints the lake in roughly 18 journeys. After that, finds turn into brass.
Why it matters. Retention falls off when the world stops changing.
Proposed refinement. Use the generative pipeline for route 2 (the city, Enzo) before launch, and add seasonal finds and pieces per season.
MediumNo way to measure it
What I found. No server means no analytics beyond App Store Connect.
Why it matters. You can't tell whether people start more, come back, or use ferries.
Proposed refinement. App Analytics plus TestFlight feedback, and the same CloudKit aggregate counts that feed the honest presence signal.
MediumAccessibility audit still flags 24–28 items
What I found. Mostly Apple's heuristics (clipping guesses, a system button, off-screen elements).
Why it matters. The test can't gate releases while it always fails.
Proposed refinement. Filter only the verified false positives in the test and keep the rest failing.
LowLaunch surface not started
What I found. Placeholder prices, no App Store page or screenshots, no Embark marketing page. The Watch screenshot here shows an older orange tint that is already fixed in code.
Why it matters. Needed before external testing.
Proposed refinement. After the High items above.
Refinements, in order
- Honest live presence + face-down audio. It makes the core promise true on day one.
- A two-device ferry test. It proves the core works.
- Fix the season pass's Enzo promise. It removes a trust and review risk.
- Simplify the start screen with progressive disclosure.
- Rethink AR as an extra, not the default.
- One painted world map.
- Route 2 content before launch.
These are proposals, not decisions. Each would become an RLY decision once you choose.