Experiments

Problem

Can generative layout feel art-directed rather than random?

Working answer

Build

• EngineZineKit, a pure Foundation and Core Graphics composition layer: no UI framework touches it, so every layout decision is testable on the host
• AnalysisVision and Core Image, run once at ingest, persisted by image hash and analysis version; never re-run on a determinism-sensitive path
• GenerationSeeded SplitMix64 randomness through a single generation context; positions, rotations and treatments quantised before a plan persists
• RenderCore Graphics and Core Text at any output scale; restore and export replay the plan, never regenerate
• DisciplineTest-first engine development; every user-visible generation persists its resolved render plan

Test

Shipped on the App Store. Restore and export replay the saved plan rather than regenerate it, and the engine was built test-first.

Learned

What it taught. Generative systems become trustworthy when surprise is separated from reproducibility. Delight lives in the generation; trust lives in the replay.

Problem

Can language learning behave like listening to an album rather than completing a lesson?

Working answer

Build

• FormatA bilingual chapter grammar with anchor lines and controlled variation: the format is the product, the model is staff
• CastingA multilingual TTS casting desk: voices audition for a part, a take once locked is never re-rendered
• ReviewNative-language review and pronunciation gates before any chapter masters; evaluation logs on every pass
• TruthStructured manifests as the source of truth for every volume and chapter; render tooling replays them deterministically
• SurfaceNext.js and Stripe storefront at driftspeak.com; native iOS listening app in App Review

Test

Thirty volumes in eight languages, each through native-language review and pronunciation gates, with evaluation logs on every pass. The storefront is live at driftspeak.com.

Learned

What it taught. The model is not the product. A durable generative-media product needs format, taste, review, and a production system that preserves all three at volume thirty as strictly as at volume one.

Problem

What should an agent remember, when yesterday's correct decision may be wrong today?

Working answer

Build

• LedgerAppend-only records; when a decision changes, a supersession chain forms rather than an overwrite
• StalenessSemantic staleness scoring as the code underneath a decision moves; contradictions flagged between builders
• RetrievalLocal vector embeddings and semantic recall; synthesis cites its sources, so the current position stays traceable to evidence
• ShapeSQLite, offline-first, git-native team sync, worktree-aware for agent fleets; exposed to any agent over MCP

Test

Released as a public MCP server. It runs as the decision ledger under the studio's own agent sessions.

Learned

What it taught. Long-term agent usefulness depends less on retaining more context than on knowing which retained context is still true. Memory without staleness detection gets more dangerous as it grows.

Problem

What should a room know about its own door, when an RSVP is a promise and not an arrival?

Working answer

Build

• DoorGuest lists, check-in from a phone, and tap cards fixed to the room; one presence record per guest per event, however they arrived
• ForecastAn expected turnout, calculated from attendance history by arithmetic; no model call anywhere
• ReturnInvitations ordered by who showed up before; reconfirmation the day before, waitlists, and held places that expire

Test

Running in production at meetcircuit.com. The forecast makes no model call, so every expected number traces back to recorded attendance.

Learned

What it taught. A system cannot record an absence it never looked for. A no-show only exists once the door has been run, so the hard part of measuring a room was never the arithmetic. It is the minute at the door.

Problem

Can learning lines feel like playing an instrument rather than rereading a document?

Working answer

Build

• StackSwiftUI and SwiftData, Swift 6 strict concurrency, Liquid Glass native
• ArchitectureBusiness rules in a fully unit-tested store; pure text-analysis functions; thin SwiftUI views
• DisciplineSwift Testing against in-memory SwiftData; XcodeGen, the project file never committed
• StanceNo backend, no accounts, no analytics: the rehearsal room is private

Test

Shipped on the App Store. Its business rules are covered by Swift Testing against in-memory SwiftData.

Learned

What it taught. A small transformation produces a new behaviour when the interaction commits fully to the user's real task. Restraint is a feature that ships.

Problem

Can a generative writing system preserve one voice, without loading the whole identity into every context window?

Working answer

Build

• RoutingThirty routes, one per kind of task; each names the files an agent reads, in order, under a context ceiling it must not exceed
• HierarchyOne file read before everything: position, precedence, what decides when rules disagree. One file read last: the refusals
• AuditsMechanical checks before model checks: grep-level tics caught by script, judgment failures caught by adversarial read
• RegressionA frozen fixture corpus; any change to the system's laws runs against it before merging

Test

In daily production as the studio's writing system. Any change to its laws runs against a frozen regression corpus before it merges.

Learned

What it taught. The system's value is not in what it can produce; any operator can get competence from a frontier model now. The value is in what it declines to produce, and in noticing when its own walls move.