Public build program | Updated June 2026
Fractalish AI
Fractalish AI is now disclosed publicly as the build-program name for the mobile activation line growing out of Fractalish, Cognitive Basin, Natural Math, AIA, and the Motorola device work.
The strongest current product-facing internal name is BasinLab. A clean split is: product BasinLab, subtitle Fractalish AI activation.
Product name, actual build conditions
BasinLab is the cleaner product surface
Fractalish AI names the public build program, but BasinLab is the name that best fits the product we are trying to bring into existence on device: local, retained, governed, replayable, and materially constrained by the handset it runs on.
The point is not to port a web demo onto Android. The point is to cultivate a native execution environment that can sense, decide, persist, replay, and recover on real hardware.
What Fractalish AI reuses cleanly
From AIA
Android shell pattern, stock-device acceptance workflow, local-only bias, device receipts, and build toolchain expectations.
From Cognitive Basin
Epistemic and action-state canon, governed ActionProposal, replayable event logs, Basin updates, narrative/review structure, and local session logic.
From Natural Math
A separate deterministic process engine with EXTEND, SENSE, and RESTRICT, event histories, parameter profiles, and formal source material.
From Fractalish
The outward morphology vocabulary and the discipline that not every visible pattern should be over-read into a claim.
What must be rebuilt for phone
- The execution backend must become mobile-native rather than desktop subprocess-oriented.
- The service surface should prefer an embedded local app architecture over desktop-style local HTTP assumptions.
- The Natural Math runtime should remain a distinct engine and can be re-authored natively instead of blindly transliterated.
- Heavy desktop acceptance harnesses should become smaller deterministic mobile traces.
Recommended mobile architecture
| Layer | Purpose |
|---|---|
| Activation Shell | Own lifecycle, local storage, review UI, replay/export, and session control. |
| Basin Core Mobile | Implement Cognitive Basin enums, governed actions, session event log, prediction/residual tracking, and replayable state reconstruction. |
| Natural Math Engine | Run distinct local-process simulations under EXTEND / SENSE / RESTRICT without collapsing into Basin governance. |
| Bridge Layer | Translate Natural Math outputs into Basin observations, anomalies, tensions, and planning signals. |
Best first phone slice
- Start an activation session with a purpose anchor.
- Run one local Natural Math simulation profile.
- Convert simulation outputs into Cognitive Basin observations.
- Produce governed state updates and next-step proposals.
- Persist replayable session history on device.
- Export a redacted trace bundle.
Why we are naming it publicly now
The work has crossed the threshold where hiding the build name is less useful than inviting capable people to inspect and help. Fractalish AI is no longer just a private aspiration. It is a real architecture-and-device program with public code surfaces, public white papers, and real implementation traction.
Public anchors
Claim boundary
Publicly naming Fractalish AI does not flatten the project's own stated caution. The site still rejects unearned claims of proven sentience, artificial personhood, or regulatory authority. The invitation is to help build and test a governed machine cognition architecture, not to bypass the proof burden.